Ambient assisted living and active aging 5th international work conference iwaal 2013 carrillo costa
Ambient
Rica December 2 6 2013 Proceedings 1st Edition Miguel Ángel Valero
Visit to download the full and correct content document: https://textbookfull.com/product/ambient-assisted-living-and-active-aging-5th-internati onal-work-conference-iwaal-2013-carrillo-costa-rica-december-2-6-2013-proceedings1st-edition-miguel-angel-valero/
More products digital (pdf, epub, mobi) instant download maybe you interests ...
Ambient Intelligence 4th International Joint Conference
AmI 2013 Dublin Ireland December 3 5 2013 Proceedings 1st Edition Serge Autexier
A Knowledge Based Framework to Support Active Aging at Home Based Environments
Miguel Ángel Valero1, José Bravo2, Juan Manuel García3, Diego López-de-Ipiña4, and Ana Gómez1
1 Dep. of Telematic Enginering and Architectures, Universidad Politécnica de Madrid, Spain {mavalero,agomez}@diatel.upm.es
2 MAmI Research Lab, Universidad de Castilla La Mancha, Ciudad Real, Spain
Jose.Bravo@uclm.es
3 Universidad de Alicante, Spain juanma@dtic.ua.es
4 MORE Lab, Universidad de Deusto, Bilbao, Spain dipina@eside.deusto.es
Abstract. Information and Communication Technologies can support Active Aging strategies in a scenario like the Smart Home. This paper details a person centered distributed framework, called TALISMAN+, whose aim is to promote personal autonomy by taking advantage of knowledge based technologies, sensors networks, mobile devices and internet. The proposed solution can support an elderly person to keep living alone at his house without being obliged to move to a residential center. The framework is composed by five subsystems: a reasoning module that is able to take local decisions at home in order to support active aging, a biomedical variables telemonitorisation platform running on a mobile device, a hybrid reasoning middleware aimed to assess cardiovascular risk in a remote way, a private vision based sensor subsystem, and a secure telematics solution that guarantees confidentiality for personal information. TALISMAN+ framework deployment is being evaluated at a real environment like the Accessible Digital Home.
Keywords: Active aging, smart home, collaborative reasoning agents, sensors.
1 Framework Contextualization
Active Aging is not just a set of recommendations for physical and psychological well-being but a “process of optimizing opportunities for health, participation and security in order to enhance quality of life as people age” [1]. The World Health Organization (WHO) highlights the necessity to promote effective strategies and solutions that maintain autonomy as a person grows older. Many daily life activities are carried out at the home environment and this user domain is widely surrounded by multiple devices and appliances which are supposed to make our life “easier”.
The interoperation at home of these available mechanic, electronic, information and communication technologies, sets the basis to provide the elderly population
C. Nugent, A. Coronato, and J. Bravo (Eds.): IWAAL 2013, LNCS 8277, pp. 1–8, 2013.
and/or people with neurodegenerative diseases with accessible services to promote personal autonomy. Cheek et al. mention the concept of aging-in-place and point out different facilities to be supported by Smart Home (SH) technologies such as “emergency care, fall prevention & detection, reminder systems and assistance for those with cognitive impairments” [2]. This idea is not new, some authors like Williams et al. described in 1998 their future smart home for “the provision of artificial intelligence -AI- based information processing and the management of decision-making structures required” [3]. Fifteen years later, users could request for SH technologies at their house since environmental sensors like presence, motion, fire, flood or gas are market available; biomedical data devices can be connected to measure pulse, temperature, glucose or blood pressure, and reasoning middleware environments are highly usable. However, the easy to use and knowledge based person centered interaction with these devices is still not solved especially to support aforementioned SH facilities. TALISMAN+, the distributed framework detailed in this research paper, aims to go one step forward in this direction by offering an integrated solution that encompasses local and remote reasoning modules, biomedical mobile connected telemonitoring devices, environmental sensor networks and security mechanisms that guarantee privacy and confidentiality of supported homecare telematic services. The framework core has been designed and developed by taking into account the necessities of two users domains: vulnerable people with Parkinson disease and persons with mobility restrictions.
Information and communication technologies can help people with cognitive and mobility impairments to promote Active Aging activities related to communication, stimulation and environmental control. Laiseca et al. showed the utilization of these technologies to assist the elderlies with cognitive disabilities by using memory games that facilitate information to caregivers and relatives [4]. Activity recognition can be triggered both from data driven information provided by users or through the utilization of sensor-based recognition. Chen et al. compare these two approaches and conclude that Knowledge-Driven models need to handle uncertainty and time in order to distinguish intent or goal recognition [5]. Ontology-based systems have been tested to support active health with mobile technologies. Docksteader et al. published a Mobile Ontology-based Reasoning and Feedback system that monitors SpO2, using Semantic Web Rule Language (SWRL) and communicate them via SMS and HTTP protocols [6]. Health care domain has also experimented with the use of cameras in private spaces in the field of Ambient-Assisted Living (AAL) and aging in place. Cardinaux et al. reviewed in 2011 the pros and contras, related to user´s acceptance, reliable reasoning and privacy, of video based technology for AAL [7].
No doubt that TALISMAN+ framework provides facilities that can be critical for the security of the elderly. Therefore, privacy of monitored individuals should be guaranteed at the same time its identities are checked as Islam et al. states for SH [8]. Since sensors and devices used can be perceived as an intrusive element at home [12], privacy and authentication issues were considered an essential part of this framework in order to reinforce the trust of users to promote their personal autonomy.
2 User Driven Design Methodology
Further to ANSI/IEEE 1471-2000 conceptual framework for architecture description, TALISMAN+ design can be decomposed in five architectural views. The methods provided by this standard helped to describe the global view of the solution according to a user driven approach. Thus, people with cognitive or mobility restrictions, such as the elderly, become the main stakeholder addressed by each of the five deployed subsystems. The mission is to support autonomous active aging at home by providing context-aware reactions triggered by detected events, user profiles and reasoning procedures. Main concerns addressed in this framework deal with security, accessibility, reliable reasoning and interoperability. Security view followed a user centered approach so that an aging person may feel trust about his or her interaction with the system. Therefore, user requirements analysis led to define a user interface so that all the stakeholders may simulate, check and effectively validate the suggestions provided by the framework. These stakeholders include the elderly person, informal caregivers, relatives, and professionals in charge such us geriatricians, therapists, social workers or nurses.
By following a user driven design methodology, replicable user cases were defined according to the knowledge acquired from two users´ entities: Madrid Parkinson´s disease association and the association of people with spinal injury and physical disabilities (ASPAYM). As Gass et al. state for internet-based services, the specification of end-user-driven data acquirement at the SH was a critical issue to define context-aware interoperable facilities to promote active aging [9].
Fig. 1 details the functional Framework design whose user driven main use case is detailed as follow: 1) Biomedical data provided by users through mobile connected sensors is sent to the reasoning subsystems; 2) the hybrid remote reasoner validates a user profile (e.g. level of risk disease) and sends this data to the local reasoner; 3) environmental context and user profile info is updated to the local reasoner; 4) a descriptive local reasoner suggests actions at the SH for active aging; 5) the security view ensures authentication, confidentiality and integrity of managed information.
Fig. 1. User-driven functional Framework Design
3 Development Results
The resulting distributed framework is described in the following parts by following the user-driven context previously depicted. :
3.1 Mobile Monitoring (MoMo) Platform
The Mobile Monitoring platform (MoMo) allows aging patients to having continuous diseases control and direct communication with their doctor. MoMo enables patient mobile telemonitoring by using biometric devices (e.g. glucometers, blood pressure meters) to send data to a mobile phone via technologies such as WiFi, NFC or Bluetooth [10]. An ontological architecture has been created in order to catalogue the elements and provide TALISMAN+ local and remote reasoners with ad hoc feedback. Patient monitoring represents one of the key elements in the progress and control of his illness. This monitoring provides patient and doctor with continuous data about disease´s status (vital signs, pulse glucose) so that, the doctor can accordingly readjust the initial treatments and prescriptions. Mobile phone is the selected technology as it is fairly used by aging people and can support daily activities for communication and information management. A group of ontologies called MoMOntology represent the ontologies of mobile monitoring process and allow to model the data collected from biometrics devices. An analytical engine, described in 3.2 which combines Fuzzy Logic and probabilistic reasoning, allows managing patient records based on an analysis of past situations to predict future difficulties like variations in vital signs). MoMo takes advantage of mobile phones and biometric devices to facilitate patient monitoring as data is recorded in a central server to be used by TALISMAN+ framework.
Fig. 2 shows the developed platform. On the left, healthcare and monitoring devices are connected to Bluetooth mobile devices. In the center, these biometric devices are linked to a mobile phone to process sensor data, manage applications, and ensure redundant connectivity via 3G and WiFi data networks. Information is transmitted to a central database and advisory system for evaluation by the medical server (right).
Fig. 2. MoMo platform and relationship between data and device entities
3.2 Hybrid Cooperative Reasoner
As stated in TALISMAN+ framework, a middleware is requested to facilitate context capture from any device, including embedded ones, to program environment’s reactivity. TALIS+Engine is the susbsystem were the ontology AMBI2ONT was created to model ambiguity in its two facets, uncertainty and vagueness, together with sensor fusion and reasoning inference engine. Upon uncertainty, the certainty factor (CF) of contextual data is modeled. Vagueness consideration allows model unclearly defined situations like cold room or noise room where different users have different perceptions. Such ontology models places, things in those places such as devices or people, capabilities and linguistic terms. Fig. 3 shows a ContextData individual with a sensor value associated a certainty degree about the credibility of such measure and a set of linguistic terms where each term is associated with a membership function. Thus, in this case the temperature sensor in a given room can be considered mainly hot. Once the measures are modeled considering the uncertainty and the vagueness of the data, semantic inference process is applied, so that implicit data is derived from explicit data. For example, the location associated to each measure is determined, knowing the location of the temperature sensor. Subsequently, a data fusion process is applied which aggregates measures of the same type (e.g. temperature sensor) within the same container (e.g. room). Two different strategies are supported: tourney (the best) and combination, where depending on the scheme applied the best, worst or average measure can be considered. Thus, the temperature values of all sensors within a room are combined.
Finally, the behavior rules defined for an extension of the JFuzzyLogic fuzzy logic engine [13] are executed. Such engine was modified to incorporate to it treatment of both uncertain data and rules. An example of the now supported syntax, which extends JFuzzyLogic engine’s Fuzzy control language (FCL), is shown on the bottom left hand side of Fig. 3, where both uncertainty (CF 1) and vagueness (HOT) are considered:
RULE CF 1
LaboratoryA hasGlobalTemperature
temperatureX temperatureX HOT -> airCoditioningX type AirConditioning airCoditioningX location LaboratoryA airCoditioningX temperature 22
Fig. 3. Hybrid reasoner uncertainty management and AMBI2ONT Ontology
3.3 Vision Based Sensor Subsystem
TALISMAN+ distributed framework relies on private sensoring solutions at the SH to make ad hoc local reasoning to promote healthy and social activities related to active aging. Vision@home is the integrated subsystem that includes a technological infrastructure and vision-based services to monitor and recognize the activity of users at home, including the privacy protection of who perceived with these vision devices. Thus, a double functionality is achieved: dynamically modeling of habitual behavior of people to detect events that may identify abnormal behaviors, and confidential vision-based services to detect and recognize objects and people of interest in the scene and characterization and interpretation of movement to monitor their activity.
Static cameras, pan-tilt, omni-directional and low cost devices RGB-D, like Kinect, were applied to allow better interpretation of motion without forgetting ethical aspects when monitoring people in private spaces. Two cases were addressed to recognize human activity: action recognition using several cameras and action recognition using Kinects. The implemented system recognizes, through cameras, simple actions such as walking, jumping, running and falling. Anonymous people silhouettes are gathered using background removal and next features invariant to scaling and rotation. During recognition, the position of the person is determined at every moment by matching the sequence of postures with the Knowledge Base using Dynamic Time Warping. The system operates at video rate, which is one of the main requirements. Regarding the recognition of actions using Kinects, a classification system was developed from skeletal actions of a person. Thus, Microsoft SDK and OpenNI library functions were tested and positions recognized by Growing Neural Gas. The optimal set of characteristics that increases the accuracy of action recognition algorithm was founded by using evolutionary learning techniques [11]. Fall detection service and other actions are supported by a simple prototype that determines the fall by calculating the hip height.
A context oriented privacy protection model was defined based on levels. Each level defines the nature of the information that will be provided to TALISMAN+ local reasoner or even shown to an authorized caregiver. Each level of representation can depend on the event and the permissions of the stakeholder. Four protection levels were verified as feasible: no alarm, which displays a virtual image of the environment without showing any person; low level alarm, which shows a virtual image of both the environment and the person, showing its location but not its posture; high level alarm, which shows a virtual image of both the environment and the person; and very high level alarm that shows a real image of the environment (Fig. 4).
Fig. 4. Vision based privacy degrees depending on the level of alarm
3.4 Local Logic Description Reasoner and Security View
The distributed framework proposed provides functionality in a home environment where critical assurance of privacy is needed for the elderly. Local reasoning should confirm individuals’ identities and ensure privacy when they are monitored. Very often, sensors and devices used for these services are perceived as an intrusive element at home. Therefore, privacy and authentication was considered an essential part of TALISMAN+ not just to protect communications but to reinforce users’ trust in the use of the news services for active aging and promotion of personal autonomy.
Bearing in mind this user-centered approach, the SH logs all interactions of sensors and monitoring systems that are managing data with the database as well as the data exchanged between the SH and telecare centers. In this way, users with cognitive or physical impairments gain access at any time to the information obtained from the logs, as the system enables tangible and understandable interaction when coupling sensors and actuators into actions. These actions allow users to mentally represent the capabilities of the SH where they perform their daily tasks, regardless of the complexity of the underlying ubiquitous system. The framework implemented a security agent in charge of sending, securing and logging the outward interactions. Fig. 5 shows this agent that receives data from SH and establishes a secure SSL channel between home and telecare entities to provide two-way authentication, non-repudiation, confidentiality and integrity services for the exchange. Previous images processing by the vision based sensor subsystem, this agent sends a virtual hidden image of people at home. For this purpose it was defined an XML register document that contents the complete sequence of interactions that take place between the server and the smart home as a result of the execution of the service. After ensuring the security of communications for the house with the outside, there is still an important point to solve referred to the way that people at home are identified. Mutual authentication in SSL requires users must be in possession of the corresponding X.509 certificate to operate the system.
Fig. 5. TALISMAN+ security supporting local reasoning engine
4 Conclusions
The deployment of TALISMAN+ framework at the Accessible Digital Home allows testing a sustainable telecare service with users from Parkinson and Cordial Injury associations. This stage will be initiated shortly once the development and integration stages are completed in order to check reliable reasoning, security and performance.
Acknowledgments. Authors would like to thank the National Plan for Science, Development and Innovation of the Spanish Ministry for Economy and Competitiveness that supported TALISMAN+ (TIN2010-20510) research results detailed in this paper.
References
1. World Health Organization: Active Ageing. A Policy Framework. Second United Nations World Assembly on Ageing, Madrid, Spain (2002)
2. Cheek, P., Nikpour, L., Nowlin, H.D.: Aging Well With Smart Technology. Nursing Administration Quarterly 29(4), 329–338 (2005)
3. Williams, G., Doughty, K., Bradley, D.A.: A Systems Approach to Achieving CarerNet— An Integrated and Intelligent Telecare System. IEEE Transactions on Information Technology in Biomedicine 2(1), 1–9 (1998)
4. Laiseca, X., Castillejo, E., Orduña, P., Gómez-Goiri, A., López-de-Ipiña, D., González Aguado, E.: Distributed Tracking System for Patients with Cognitive Impairments. In: Bravo, J., Hervás, R., Villarreal, V. (eds.) IWAAL 2011. LNCS, vol. 6693, pp. 49–56. Springer, Heidelberg (2011)
5. Chen, L., Hoey, J., Nugent, C.D., Cook, D.J., Yu, Z.: Sensor-based Activity Recognition. IEEE Transactions on Systems, Man, and Cybernetics, Part C: Applications and Reviews 42(6), 790–808 (2012)
6. Docksteader, L., Benlamri, R.: MORF: A Mobile Health-Monitoring Platform. IT Professional 12(3), 18–25 (2010)
7. Cardinaux, F., Bhowmik, D., Abhayaratne, C., Hawley, M.S.: Video based technology for ambient assisted living: A review of the literature. Journal of Ambient Intelligent Smart Environments 3(3), 253–269 (2011)
8. Islam, K., Shen, W., Wang, X.: Security and Privacy Considerations for Wireless Sensor Networks in Smart Home Environments. In: Proceedings of the IEEE 16th International Conference on Computer Supported Cooperative Work in Design, pp. 626–633 (2012)
9. Gass, O., Maedche, A.: Enabling End-user-driven Data Interoperability – A Design Science Research Project. AMCIS 2011 Proceedings - All Submissions, paper 221 (2011)
10. Villarreal, V., Bravo, J., et al.: Towards Ubiquitous Mobile Monitoring for Health-care and Ambient Assisted Living. In: International Workshop on Ambient Assisted Living, Valencia, Spain (2010)
11. Chaaraoui, A., Climent, P., Flórez, F.: A Review on Vision Techniques applied to Human Behaviour Analysis for Ambient–Assisted Living. Expert Systems with Applications (in press), http://dx.doi.org/10.1016/j.eswa.2012.03.005
12. Balta-Ozkan, N., et al.: Social barriers to the adoption of smart homes. Energy Policy (2013), http://dx.doi.org/10.1016/j.enpol.2013.08.043
13. Cingolani, P., Alcalá-Fdez, J.: jFuzzyLogic: a Java Library to Design Fuzzy Logic Controllers According to the Standard for Fuzzy Control Programming. International Journal of Computational Intelligence Systems 6 (supp. 1) (2013)
Mobile Based Prompted Labeling of Large Scale Activity Data
Ian Cleland1, Manhyung Han2, Christopher Nugent1, Hosung Lee2, Shuai Zhang1, Sally McClean3, and Sungyoung Lee2
1 Computer Science Research Institute and School of Computing and Mathematics, University of Ulster, Newtownabbey, Co. Antrim, Northern Ireland, BT37 0QB {i.cleland,cd.nugent,s.zhang}@ulster.ac.uk
2 Dept. of Computer Engineering, Kyung Hee University, Korea {smiley,hslee,sylee}@oslab.khu.ac.kr
3 Computer Science Research Institute and School of Computing and Information Engineering, University of Ulster, Coleraine, Northern Ireland, BT52 1SA si.mcclean@ulster.ac.uk
Abstract. This paper describes the use of a prompted labeling solution to obtain class labels for user activity and context information on a mobile device. Based on the output from an activity recognition module, the prompt labeling module polls for class transitions from any of the activities (e.g. walking, running) to the standing still activity. Once a transition has been detected the system prompts the user, through the provision of a message on the mobile phone, to provide a label for the last activity that was carried out. This label, along with the raw sensor data is then stored locally prior to being uploaded to cloud storage. The paper provides technical details of how and when the system prompts the user for an activity label and discusses the information that can be gleaned from sensor data. This system allows for activity and context information to be collected on a large scale. Data can then be used within new opportunities in data mining and modeling of user context for a variety of applications.
1 Introduction
The ubiquitous nature of smart phones within our everyday lives provides new opportunities to collect real time context information, such as activity, location and social interactions, from a large number of users [1]. This large amount of data has the potential to be used in a number of application areas such as activity promotion, self management of long term chronic health conditions, context aware services and life logging [2]. The automatic recognition of activities is performed through the application of machine learning techniques to data gleaned from low level sensors [3]. The training of these algorithms, from a data driven perspective, relies largely on the gathering, pre-processing, segmentation and annotation of the sensor data into distinct classes [4]. The data must therefore be correctly labeled prior to being used as a training set in a machine learning paradigm.
C. Nugent, A. Coronato, and J. Bravo (Eds.): IWAAL 2013, LNCS 8277, pp. 9–17, 2013.
Collecting this data from a larger population under free living conditions may have the potential to improve the generalization abilities of any activity recognition (AR) models developed through provision of a larger quantity of representative data for training purposes. Such data sets should include data from a variety of sensors, recorded during a wide range of activities and contexts from a large number of users, over an extended period of time (months or years). Most importantly the data should also include accurate ground truth labels that represent user activities [5].
The use of smart phones can be viewed as one possible manner in which this large amount of data may be captured unobtrusively. Many handsets now have a range of in-built sensors, large memory storage, fast processing and low power communications, which meet the requirements of the range of data to be collected [6]. Furthermore, unlike many devices used as part of a research study, many potential subjects already own mobile phones, are accustomed to carrying them and always keep them charged [1]. Unfortunately, using mobile devices to gather data on a large scale can also prove difficult. In particular the integrity of the user annotation can be questionable. For example, users may forget to label a section of valuable data or may complete the labeling inaccurately. Nevertheless, a large scale fully annotated data set is recognised as being the key step to improve and increase the widespread adoption of AR applications [1], [6].
This paper presents an overview of a mobile based prompted labeling application aimed at overcoming the challenges associated with collecting annotated activity data on a large scale. In order to set the context of this work, a brief review of related material is presented. Following this the system architecture of the proposed prompt labeling application is described and the paper concludes with a discussion of the data which can be collected and analyzed.
2 Background
Although a large amount of research has focused on the ability to accurately detect a range of physical activities, very few studies have provided a detailed description of how the ground truth of data sets, for the purposes of a data driven approach to AR, have been acquired. To date the majority of AR studies have used data collected under structured or semi structured conditions, from a small number of participants (1-20 subjects). Participants often perform a set of preplanned tasks which are completed within a controlled environment [7], [8], [9], [10]. In this case, the ground truth is often recorded by a human observer and sensor data are then annotated offline according to the observer. This is viewed as being necessary as it allows researchers to capture the ground truth, when labeling data, in an effort to create highly accurate data sets. Data collected in this manner may not, however, be truly representative of completing the task in a free living environment. Furthermore, labeling and processing data is this manner can be a laborious and time consuming task for researchers. Boa and Intillie asked participants to complete a list of planned activities and to note the time at which they started and completed each activity [8]. Again this process of continuously noting the time at which an activity is commenced and
completed is fine for short term laboratory based studies, however, would not be feasible in the long term under free living conditions.
In order to allow the collection of data in a more free living environment, researchers have utilized video cameras [11]. The subsequent video recording is reviewed offline to identify what activity was being performed at a particular time. Similar techniques have been used within smart environments to label the onset/ completion of object interactions [12]. Using groups of labelers sourced from the crowd is viewed as one way of dealing with the labour intensity of the task. Lasecki et al. [13] used activity labels, generated by groups of crowd sourced labelers to annotate activities from video data. All of the aforementioned labeling methods are labor intensive and time consuming and some approaches, in particular video annotation, can have implications with data privacy. Furthermore, the need to install or issue video cameras for recoding the activities reduces the scalability of the approach.
Alternatively on a larger scale, users are often asked to annotate their own data using a mobile interface. This usually requires the user to start and stop the data capture process manually [14]. When using the application the user is then asked to label the activity they have just or are about to complete. Although this method is relatively accurate at segmenting the activity it requires the user to explicitly start and stop recording. Tapia et al. [15] used a technique based on the experience sampling method to trigger self reported diary entries every 15 minutes. Multiple choice questions were answered by the user to select which of the 35 activities users were completing. Due to the intermittent nature of the labels it was found to be difficult to detect short time activities. The process of continually labeling can become laborious for users, particularly when performed over an extended period of time. Furthermore, this can result in the user providing incorrect labels for the data or simply not engaging with the system at all. In order for the user to input a label, some interaction with the mobile device is required. This may interrupt the user during the activity, which in turn may impact upon the activity that the person is undertaking, thus impacting overall on the data recorded. In an attempt to address the issue of interaction voice recognition has been used for the purposes of annotation [16]. The mobile device listens for key words such as "start activity" and "stop activity" to start and stop the recording. Voice recognition is then used to label the activity, with the user again saying keywords, such as "standing" or "walking". Nevertheless, having the smart phone continuously listening for keywords can consume battery power and may hamper the usability of the application. Additionally, inaccuracies of voice recognition can lead to mislabeling of data.
Our approach uses prompted labeling, driven by an underlying mobile based AR module, in an effort to improve the process of collecting and annotating data. Users can annotate their everyday activities through use of a personalized mobile application. When the user is detected as standing still, a prompt is provided to enable the user to label the activity they were previously completing. In this manner the sensor data for the respective activity is segmented and saved automatically and an activity label is supplied by the user after the activity has been finished thus maintaining the integrity of the data.
3 Prompted Data Labeling
The proposed mobile application is based upon the principle of prompts to label a user's context and activity data. At periodic times throughout the day, the application will prompt the user to confirm which activities they have just completed. In addition to user reported data, additional information gleaned from the mobile device, such as automated activity classifications, GPS latitude and longitude, accelerometry data and bluetooth interactions will also be recorded. This additional data will aid in further contextualizing the annotated data sets with the intention of improving the validity of labelling. An overview of the system architecture of the proposed application is shown in Fig. 1.
Fig. 1. An overview of personalized mobile application for prompt labelling. The prompted labeling module sits on top of an existing activity recognition module and periodically prompts users to label their activity. The architecture includes mobile services to support the secure transmission and processing of data in addition to the collection of other sensory data available from the mobile platform.
In order to enhance user engagement and compliance of the application it is important that the prompted labeling module is to be incorporated within an application which provides some incentive through appropriate feedback mechanisms. This type of application could include any context aware application such as an activity monitor, calorie counter or context aware services. A suite of mobile services will be developed to ensure the secure processing and transmission of all data collected from the users. These services will be responsible for managing security, efficient transmission of data and interfacing with cloud services. A brief description of these components is provided in Table 1.
Table 1. Provides a description of the main components of the system architecture
Component Name
Prompt labeling module
Activity recognition module
Data management
Cloud services
Context aware application
Component Description
This component contains a splash screen which allows the user to select a label for their activity data from a predefined list.
The activity recognition module attempts to detect changes in activity class to the standing still activity. From this a prompt is then initiated.
This module ensures the data is appropriately structured and formatted to ensure efficient transfer and storage. In this respect the security of the sensitive data is crucial, therefore efficient cryptography protocols shall be employed.
Cloud services provide the appropriate infrastructure to support data storage analysis and mining of the large data set.
The prompter sits upon a context aware application which enhances user engagement by providing tailored feedback (e.g. activity levels, calorie counting and context aware services)
3.1 Activity Recognition Module
The AR model used within this work, originally developed by Han et al. [17], utilizes multimodal sensor data from accelerometery, audio, GPS and Wi-Fi to classify a range of everyday activities such as walking, jogging and using transport. The activity recognition is performed independently of the position or orientation of the smart phone. This approach increases the practicality and usability of the system as the phone can be carried at any location and the AR is not affected by the user interacting with the device. Data from the accelerometer is used to detect transitions between ambulatory activities to activities which involve the use of transport. Accelerometer data, sampled at 50Hz, is computed into time and frequency domain features which are used as inputs to a Gaussian Mixture Classifier. Audio data is used in the classification if there is a need to classify between transportation activities (riding a bus or subway). Only using the audio when necessary allows the power consumption on the Smartphone to be minimized. GPS and Wi-Fi signals are then used to validate the classification between activities. Speed information, derived from GPS is used to determine whether a user is walking, running or standing still. The Wi-Fi signal is used to differentiate between bus and subway activities, as very few public or private wireless networks are available within the subway system.
3.2 Prompted Labeling Module
The prompted labeling module (PLM) prompts the user to label the activity they have just completed. Based on the output from the AR module the PLM polls for class
transitions from any of the activities (e.g. walking and running) to the standing still activity. Once a transition has been detected the PLM prompts the user, through the provision of a message on the mobile phone, to provide a label for the last activity that was carried out. The raw data from the accelerometry sensors are then stored to the mobile device before being transmitted to the cloud for storage and further processing. By prompting the user to label the activity it is possible to verify that the activity has been correctly identified by the AR module. In this way the trustworthiness of the AR module can be validated in addition to providing a fully annotated data set. Fig. 2 presents an example of interaction with the prompt labeling screen on the mobile device.
Fig. 2. An example of user interaction with the prompt labeling screen. The AR module detects a change in class from the original activity to standing still. The prompt is then issued for the user to label the previous activity. Raw sensor data, for this activity, is then saved to the mobile device before being uploaded to the cloud for further processing and storage.
Fig. 3. Illustrates how activities are detected from the raw accelerometer signal by the AR module. Activates are detected every 3 seconds, three consecutive detections are used to label the activity. The prompt is initiated when the AR module detects a change in class from one activity to standing still.
The AR module detects an activity based on three seconds (150 samples) of data. Three consecutive detections (9 seconds) are then used to label the activity. This is carried out in order to limit the number of detection errors. Once the AR module
detects a change from the current activity to the standing still activity for 9 seconds the previous activity data from the sensors is saved to memory. This process, from the perspective of raw accelerometry data is shown in Fig. 3. Currently, the prompt is initiated every time the AR detects a transition from an activity to standing still. It is envisioned that when the application is rolled out on a larger scale the user will be able to set how many prompts they receive per day in order to improve adoption of the system.
Currently data recorded by the system is stored directly to the local memory of the Smartphone, in the form of a text file. Fig. 4 shows the structure of this file. Data includes date and time stamp, raw accelerometer values (X, Y and Z axis), GPS latitude and longitude in addition to the Class label from the AR module and the prompted user label (named AR Label and User Label respectively). It is envisioned that this data could then be encrypted before being transmitted and stored in the cloud.
Fig. 4. Shows an example of data recorded by the prompted labeling module. Recorded data includes, Date and time stamp, sample number, Accelerometer data from each axis, GPS latitude and longitude, in addition to the Class label from the AR module and prompted user label.
4 Summary
The ability to collect contextual information, such as activity, location or social interactions, on a large scale is becoming increasingly important. Such data sets allow for a deeper understanding of a population's activity habits and allow information to be delivered in a context sensitive manner. Current methods of collecting contextual information, particularly activity data, are normally limited to small scale studies. This is partly due to issues surrounding the ability to obtain ground truth information to annotate such data. The current approach aims to address such issues, through the use of a mobile based context aware PLM which prompts the user to supply label
information for their current activities. In turn this improves the validity of data labels, which can then be used to improve the accuracy of data driven activity models. Transmitting and storing this data within the cloud opens new possibilities to exploit cloud services in order to mine these big data sets further in order to provide a deeper understanding of activity trends within healthcare. Plans for future work involve the evaluation of the current solution in order to assess its ability to accurately label data in a free living environment.
Acknowledgement. The authors wish to acknowledge support from the EPSRC through the MATCH programme (EP/F063822/1 and EP/G012393/1). The views expressed are those of the authors alone.
References
1. Intille, S.S., Lester, J., Sallis, J.F., et al.: New Horizons in Sensor Development. Medicine & Science in Sports & Exercise 44, S24–S31 (2012)
2. Hamm, J., Stone, B., Belkin, M., Dennis, S.: Automatic Annotation of Daily Activity from Smartphone-Based Multisensory Streams. In: Uhler, D., Mehta, K., Wong, J.L. (eds.) MobiCASE 2012. LNICST, vol. 110, pp. 328–342. Springer, Heidelberg (2013)
3. Preece, S.J., Goulermas, J.Y., Kenney, L.P.J., et al.: Activity Identification using BodyMounted sensors—a Review of Classification Techniques. Physiol. Meas. 30, R1–R33 (2009)
4. Avci, A., Bosch, S., Marin-Perianu, M., et al.: Activity Recognition using Inertial Sensing for Healthcare, Wellbeing and Sports Applications: A Survey, pp. 1–10 (2010)
5. Hossmann, T., Efstratiou, C., Mascolo, C.: Collecting Big Datasets of Human Activity One Checkin at a Time, pp. 15–20 (2012)
6. Lane, N.D., Miluzzo, E., Lu, H., et al.: A Survey of Mobile Phone Sensing. IEEE Communications Magazine 48, 140–150 (2010)
7. Krishnan, N.C., Colbry, D., Juillard, C., et al.: Real Time Human Activity Recognition using Tri-Axial Accelerometers, pp. 1–5 (2008)
8. Bao, L., Intille, S.S.: Activity Recognition from User-Annotated Acceleration Data. In: Ferscha, A., Mattern, F. (eds.) PERVASIVE 2004. LNCS, vol. 3001, pp. 1–17. Springer, Heidelberg (2004)
9. Parkka, J., Ermes, M., Korpipaa, P., et al.: Activity Classification using Realistic Data from Wearable Sensors. IEEE Transactions on Information Technology in Biomedicine 10, 119–128 (2006)
10. Mannini, A., Intille, S.S., Rosenberger, M., et al.: Activity Recognition using a Single Accelerometer Placed at the Wrist Or Ankle. Med. Sci. Sports Exerc. (2013); E-Published ahead of Print
11. Plotz, T., Chen, C., Hammerla, N.Y., et al.: Automatic Synchronization of Wearable Sensors and Video-Cameras for Ground Truth Annotation–A Practical Approach, pp. 100–103 (2012)
12. Cruciani, F., Donnelly, M.P., Nugent, C.D., Parente, G., Paggetti, C., Burns, W.: DANTE: A video based annotation tool for smart environments. In: Par, G., Morrow, P. (eds.) SCUBE 2010. LNICST, vol. 57, pp. 179–188. Springer, Heidelberg (2011)
13. Lasecki, W.S., Song, Y.C., Kautz, H., et al.: Real-Time Crowd Labeling for Deployable Activity Recognition, pp. 1203–1212 (2013)
14. Kawaguchi, N., Watanabe, H., Yang, T., et al.: HASC2012corpus: Large Scale Human Activity Corpus and its Application (2012)
15. Tapia, E.M., Intille, S.S., Larson, K.: Activity recognition in the home using simple and ubiquitous sensors. In: Ferscha, A., Mattern, F. (eds.) PERVASIVE 2004. LNCS, vol. 3001, pp. 158–175. Springer, Heidelberg (2004)
16. Harada, S., Lester, J., Patel, K., et al.: VoiceLabel: Using Speech to Label Mobile Sensor Data, pp. 69–76 (2008)
17. Han, M., Lee, Y., Lee, S.: Comprehensive Context Recognizer Based on Multimodal Sensors in a Smartphone. Sensors 12, 12588–12605 (2012)
Another random document with no related content on Scribd:
it be remembered that the fleet when at Delos must have heard of Mardonius’ retirement from Attica, and had also, it would seem, heard of the march of the home army into Bœotia, their fear strongly supports the conjecture that they knew that the object of that army was not to merely guard the passes of Kithæron, but to take the offensive against the Persian. It is hardly conceivable that, after Mardonius’ retirement, they should have felt apprehensions of this nature about the army, had its object been merely the defence of the Kithæron range, since the enemy had plainly shown by his retreat from Athens that his main desire was to be north of that range, and that he would have but little motive for throwing his men against mere defenders of passes which he had shown that he did not want to use. Furthermore, the possible intentions of Mardonius in making Thebes his headquarters may have alarmed the Greeks by indicating the nature of a policy, not indeed so terrible as that wherewith the expedition was originally undertaken, but constituting, all the same, a most serious danger to all the states of Southern Greece.
Relieved of this fear, they went into battle with a good heart, feeling, says Herodotus, “that the Hellespont and the Islands would be the prize of victory.” Not a word of the Ionian towns on the mainland. The Persian position was indeed too strong in respect to these.
H. ix. 102.
The march was evidently parallel with the shore, along it and the level ground in its neighbourhood, and also along the hillside, cut up, as the hillsides of that land always are, by numerous water-courses. The result was that the Athenians and those with them, advancing along the unimpeded level, came into contact with the enemy before the Lacedæmonians, who were advancing along the slope of the mountain. If Herodotus’ description is strictly worded, the Lacedæmonians were engaged in some kind of a turning movement.
H. ix. 102.
The Athenians were therefore probably guilty of a tactical error in beginning the attack before the movement was complete.
The Persians adopted the same form of defence as in the last fight at Platæa, using their shields as a breastwork; and for some time the battle was fought without advantage to either side, until the Athenians and those with them, eager to win the victory before the Lacedæmonians came up, broke through this barrier and fell in a mass upon the Persians, who, after an obstinate resistance, retreated within their fortifications. The Athenians, Corinthians, Sikyonians, and Trœzenians, however, followed close on their heels, and seem to have reached the breastwork with the fugitives, so that it was captured without difficulty Except the native Persian contingent, the enemy now took to flight the former, however, resisted in scattered groups with all the bravery of a great reputation, —the Old Guard of this Asiatic Waterloo. In this combat two out of the four Persian generals fell. The Lacedæmonians came up while it was in progress, and their arrival put the coping stone on the enemy’s discomfiture. But the assailants, and especially the Sikyonians, paid dearly for their victory, so Herodotus says; and Greek historians are not in the habit of exaggerating Greek losses in battles with barbarians. Meanwhile the disarmed Samians did what they could to help the Greeks, and, following their lead, the other Ionians attacked the Persians in the camp. The Milesians, who had been sent to the peaks of Mykale, now played the part which, under the circumstances, they might be expected to play. So far from acting as guides to safety, they led the routed fugitives into the very hands of the Greeks, and themselves took part with zeal in the slaughter which ensued. At last Miletus had the opportunity of avenging its destruction at the end of the Ionian revolt, and it is certain that the bitterness of the last fourteen years found ample expression and ample satisfaction on this day of revenge.
H. ix. 104, ad fin.
“Thus,” says Herodotus, “Ionia revolted a second time from the Persians.”
THE FATE OF IONIA.
He makes no comment upon the fact. It would have been difficult for him to do so. He had regarded the first revolt as a conspicuous blunder, probably because he looked upon it as premature. Even now he sings no pæan of emancipation. Maybe the Halikarnassian believed that, in
spite of the brilliant success of the year, the permanent freedom of the Ionian towns could never be secured so long as Persia retained its hold on West Asia. He wrote with well-nigh a century of experience behind him; he died a century before there arose a power upon the Ægean whose unity of strength made it fit to cope with the dead weight of the Persian power in Asia.
The Athenians were credited by popular tradition with having played the greatest part in the battle; but it may be suspected that had Leutychides’ version of the fighting survived, it might have contained some criticism of the error of a premature attack, to which the escape of a large part of the Persian army was probably due.
After destroying the enemy’s camp and fleet the Greeks withdrew to Samos with such booty as they had captured.
H. ix. 106.
The action of the Ionians at Mykale had raised the whole question of the future of the Asiatic Greeks. There was a large section in the fleet who saw,—what the sequel proved to be correct,—that the European Greeks were not in a position to maintain the freedom of continental Ionia against the Persians for an indefinite space of time. That would have meant the maintenance of a large garrison within the various towns, such as the highly composite Greek alliance could not regard as within the realm of practical politics. It was consequently proposed to tranship the Ionians en masse to the European shore, and to leave Ionia to the Persian. The strategic position of the Ionian cities was indeed fatally weak.
But, after all, it would hardly be the strategical question that presented itself to the Greeks gathered at Samos. What they did know was that these Ionian cities had failed to resist conquest by Lydia and Persia, and had failed in one tremendous effort to throw off the Persian yoke; and that one of the conspicuous causes of failure had been the difficulty of sustained united action. Of the commercial advantages and their causes, they were probably as well aware as the best instructed modern student with the best maps at his disposal.
H. ix. 106.
It was, no doubt, this balance of advantage and disadvantage which led to the division of opinion in the council as to the best course to be pursued. The Peloponnesians were in favour of planting the Ionians on the lands of the medized Greeks at home. To this proposal the Athenians offered an uncompromising resistance, and claimed that they alone had the right to decide the matter, since the Ionians were colonists of their own. The claim was a shadowy one; but the resistance was successful. It is noticeable, however, that the Asiatic Greeks who were in the first instance received into the alliance were Samians, Chians, Lesbians, and the other islanders, but none of the inhabitants of the cities of the mainland. Thus were sown the seeds of that famous Delian League which was destined in the near future to play so great a part in the history of the fifth century.
Throughout Greek history the Hellespontine region is the link between Europe and Asia. To it the Oriental seeking to win empire in Europe ever turned, while the European, whether on the defensive or the offensive against the power of the East, was at all times anxious to secure its possession. And thither now the fleet from Samos sailed, under the impression, says Herodotus, that it would find the bridge in existence. If such an impression did actually exist with the Greeks at this time, the fact bears strong evidence to the silence and desertion which in these years brooded over that highway of the nations—the Ægean. The fleet was delayed on the voyage by adverse winds, and was obliged to anchor for a time at Lektos Thence it sailed to Abydos, where the Greeks discovered for the first time that the bridge was no longer in existence.
H. ix. 114.
Hitherto Leutychides and the Spartans with him had shown in this expedition an enterprise peculiarly foreign to them; but now once more a fatal national characteristic began to reassert itself. Whether because of that homesick conservatism of the race which made it averse to ventures far beyond its borders, and anxious when engaged in them to get quit of the matter in hand at the earliest possible opportunity, or from a lack of
THE FLEET SAILS TO THE HELLESPONT.
intelligence which failed to grasp the proper issue of a situation, Sparta was ever wont to leave her tasks unfinished, especially if they demanded absence far from home. And so it was now. On the plea that the only object of the expedition to the Hellespont had been to ensure the destruction of the bridge, they renounced all idea of further operations for the time being, and set sail for home. It may be doubted whether they genuinely believed that the disappearance of the bridge had removed all necessity for further action in this region. The Athenians, at any rate, under their leader Xanthippos, saw that the peril from Persia must ever be recurrent, if that power continued to hold the Thracian Chersonese, that tongue of land which in both ancient and modern times, though for different reasons, has ever been of the greatest strategical importance to Mediterranean powers. They determined therefore to clear the enemy out of this, their tête-du-pont in Europe.
The chief strategic position in the peninsula at that time was Sestos. It lay on the European side of the great ferry of the Hellespont at Abydos, and so commanded the main route from Asia to Europe.
As a place of great military importance, it was the strongest fortress in those parts; and on the news of the arrival of the Greeks at the Hellespont, the Persian population from all the neighbouring towns collected thither under the command of Artauktes, the governor of the region. He was a man who had got an evil reputation among the Greeks owing to sacrilegious behaviour of the grossest character, peculiarly calculated to arouse their most fierce resentment. The arrival of the Greeks in the Hellespont had been so unexpected that he was taken unawares, having made no preparations for a siege. But the inexperience and incapacity of the Athenian assailants in the attack on walled towns, despite the reputation they enjoyed among their fellow-countrymen, who were themselves hopelessly impotent in this department of the art of war, caused the siege to drag out a weary length, until the wane of autumn brought with it that severity of wintry weather from which all the lands within reach of the inhospitable Euxine suffer at that time of year.
The Athenian soldiers and sailors began to weary of the apparently endless blockade, and demanded of their leaders that they should be taken home once more. This request Xanthippos and his captains refused, saying that they must bide where they were until either the town were taken or the Athenian Government sent for them. So they continued to bear their hardships as best they could. Meanwhile the besieged, reduced to the last extremity of want, ate even the leathern straps of their beds. At last, in desperation, the Persian portion of the inmates of the town escaped by night through the besiegers’ lines at a point where there were but few men on guard; but here their success ended. The natives of the Chersonese who were in the town informed the Athenians by signal of their flight, and the latter started in hot pursuit. One body of the fugitives made its escape to Thrace, to meet with a miserable fate at the hands of the wild tribes of that region; but the main body, under Artauktes, was overtaken by the Athenians near Ægospotami, and after an obstinate defence, such as survived, including Artauktes, were brought back as prisoners to Sestos. A tale which Herodotus tells shows that the Persian commander had some apprehension with regard to the retribution which his sacrilegious conduct might bring upon him; for he made an offer to Xanthippos to pay one hundred talents’ compensation for the outrage, and two hundred talents’ ransom for himself and his son. To this offer Xanthippos turned a deaf ear.
The punishment inflicted on the Persian was a blot on Greek civilization. He was nailed to a board, and his son was stoned to death before his eyes. The Greek nature was capable in moments of revenge of inflicting the death penalty in wholesale fashion on enemies who had excited its bitter resentment; but the torture of a captured foe was wholly alien to the spirit of the people. Doubtless the Persians, with the ineradicable cruelty of the Oriental, had given many precedents for such a form of revenge; but whether that be so or not, this particular act was inexcusable in a people who claimed for themselves a standard of civilization infinitely higher than that of the world around them.
END OF THE CAMPAIGN.
With the capture of Sestos the campaign of this famous year ended, and with it the great war for the liberation of European Greece. Many years were indeed fated to pass before the present struggle ceased. But from this time forward the war entered upon a new phase, in which the Greek was the assailant. Hitherto he had been acting purely on the defensive; even the expedition across the Ægean had been but an act of the great drama which was being played in Greece. The West had triumphed over the East in one great effort, wherein the success had been rapid and striking. But henceforward the tide of success was destined to flow more slowly, —so slowly, indeed, that ere the end came, victor and vanquished alike had sunk into decay, and alike had fallen into subjection to that newer, broader, and more vigorous but less genuine Hellenism which Macedonia evolved from her heritage in the older type.
CHAPTER XIII. THE WAR AS A WHOLE.
I is often an invidious task to examine the causes which lie behind any great series of events in military history, because the most efficient of them are in the majority of cases due to human error rather than to human power The historian of peaceful development is not exempt from the necessity of compiling such records of frailty, but by the very nature of things such tasks fall more frequently to one who narrates the story of periods at which the sanity of the sanest is troubled by the nervous tension involved in participation in events whose issues and their results lie in the immediate, not in the distant future. The errors of war may not be greater, but they are less remediable than those of peace. There may be time to arrest the latter’s slow decay; but the swift and often fatal stab of a lost battle is the work of a moment. Men have in all ages been conscious of the gravity of this truth; and the consciousness of it, apart altogether from the inevitable physical fear, has rendered them less capable at such times of the calculations of reason, or even, in some instances, of common sense.
The great Persian War was of a special type. In the majority of cases in which races and empires have come into collision, each side has had some practical acquaintance with the resources, devices, and fighting qualities of the other; and in many cases such experience has been intimate and prolonged. But when the Persian and the Greek of Europe came into collision in 480, such experience can hardly be said to have existed on either side; or, in so far as it did exist, it had been misleading. In only two instances had the European Greek come into contact with the Persian on the field of battle, and in both of them the same Greek state, the Athenian, had alone been represented in the conflict. But, furthermore, the
UNDETERMINED FACTORS.
instances had, from a military point of view, been indecisive, if not actually fallacious. At Ephesus, in the first year of the Ionian revolt, a small contingent of Athenians had been present on the defeated side, when the Persians fell on the expedition which had burnt Sardes. Of the battle practically nothing is recorded save the result; but this much may be assumed with certainty,—that a fight in which a small body of European Greeks had been defeated in partnership with hastily raised levies of Ionians could not possibly afford any experience worth calling such to either of the sides who were destined to take part in the war of twenty years later The Ionians of Asia, long under Persian rule, must have been very deficient in military training when compared with the Greek of Europe. Darius had, indeed, made use of them in the Scythian expedition, but in that instance had employed them to guard the lines of communication. It is manifestly improbable that the Persian Government would encourage any advanced system of military exercises in a people whose position at the very borders of the empire would have rendered them, if accustomed to the use of arms, very dangerous subjects.
Marathon was the other instance. It is a battle of problems, a problem among battles, whose data are woefully imperfect. The only thing about it which seems clear is that, for some reason which can only be conjectured, it formed but an imperfect test of the fighting capacity and methods of Greek and Persian respectively.
Thus, for all practical purposes, when the two races came into conflict in 480 they were, militarily speaking, unknown quantities to one another, and each had to learn how best to meet the strategy and tactics of the enemy The consequence was inevitable. Both sides made grave mistakes of commission and omission; and it may be even said that the victors made more than the vanquished, though not of such a fatal character. The Corinthians were not indeed wrong when in later days they summed up the causes of the issue of the war by saying that “the Persian was the rock on which he himself made shipwreck;” for of the two main reasons which led to the final victory of the Greeks, one was undoubtedly the fatal nature of the mistakes which the Persians made. They seem, relying
on their prestige, and on their enormous numbers, to have held their adversaries too cheaply; and from this fundamental error all their other errors were generated.
Knowledge after the event renders it very difficult to appreciate the circumstances of any particular date in history, and the difficulty is all the greater if the events of the time immediately succeeding be of such a nature as to completely change those circumstances. Such is markedly the case in the instance under consideration. The mental attitude of the Persian towards the Greek power after 479 is well known to the modern world; and it is recognized indeed that a great change must have been brought about in it by the events of that and the preceding year; but the extent of the change is not perhaps realized, because the very real nature of the grounds of confidence with which Persia entered upon the war are not sufficiently taken into consideration, and the efforts made by Herodotus to bring this confidence into prominence are too apt to be regarded as aiming at the greater glorification of his own race.
And yet the grounds for that confidence are plain. In a long and almost unbroken series of wars the Persian had conquered Western Asia. He had never met with the race which could face his own upon the set field of battle, and this, not in an experience of a few years, but of half a century. He might indeed feel that he had been tried in the balance of warfare and not found wanting. Man’s success against him had never been more than temporary.
It now remains to be considered why this confidence was illgrounded.
The Persian had never seen the Greek heavy-armed infantryman at his best, well disciplined, and fighting on ground suited to his tactics, save perhaps at Marathon, when the test was probably regarded as unconvincing. Herodotus points out the superiority of his panoply to that of the comparatively light-armed Persian. There is much more of the empirical than of the scientific in the lessons of war; and the experience of all ages points to the fact that an army which enjoys a noticeable superiority over its enemy in respect to
ARMS AND THE LAND.
weapons will in all probability, if other things be equal, come off victorious. Such exceptions as history can adduce to this rule are rather apparent than real, and are, in the vast majority of cases, due to the possessor of the superior arms adopting tactics either unstated to them, or wholly at variance with the nature of the region wherein fighting is being carried on. The Greeks at Platæa made a mistake of the latter kind, which was only annulled by a greater mistake made subsequently by the other side.
In this great war, then, the two most efficient causes of its issue were (1) the undue confidence of the Persian, giving rise to fatal mistakes; (2) the great superiority of the Greek panoply. A third, of a negative character, may perhaps be added, namely, that the nature of the country did not permit of the invader making use of his most formidable arm,—the cavalry. The second and third causes may be included in that wider generalization which has been already discussed;—the West on its own ground must have prevailed over the East.
The immediate preliminaries of the actual invasion of Greek territory bring into prominence the high state of efficiency which must have been attained in the military organization of the Persian empire. Leaving out of consideration the difficulties to be overcome before the huge mixed force was collected at Sardes (which town became for the time being, in place of Susa, the prime military base of the empire), the organization which enabled this great army to be brought without accident, or, in so far as present knowledge goes, without a hitch of any kind, over the eight hundred miles of difficult country which separated its base from Middle Greece, must have been the outcome of a highly effective and highly elaborated system evolved by a people whose experience was indeed large and long, but who must also have been gifted with that very high form of mental capacity which is able to carry out a great work of this nature. The secret of success,—it may almost be said of possibility,—in the present instance was the employment of the fleet for commissariat purposes. It was a method of advance not new to Persian campaigning, the first instance of its employment going back as far as the time of the invasion of Egypt by Cambyses.
On the Greek side organization on so huge a scale was not called for. Numbers were smaller, and distances in comparison insignificant; and only at the end of the campaign, when the Athenian fleet was engaged in the blockade of Sestos, had a long period of absence from the commissariat base to be provided for. Still, what was done must have been done on a system; and the system cannot be supposed to have been a bad one, for there is no hint of its ever having broken down. It was in all probability most severely tried when it became necessary to keep up the supplies of the great host during the weeks it remained at Platæa. Such mention of it as occurs in Herodotus’ narrative is of a purely incidental character. It was a side of the war in which he was not likely to take any great interest, even if he could have obtained much information on the subject, or have appreciated the significance of such information as he did obtain. The disaster to the commissariat train in the pass of Dryoskephalæ is the most prominent of the rare instances in which Herodotus mentions anything connected with this department of the service. His reference to a signalling system extending through Eubœa at the time of Artemisium,—a line of communication which was in all probability carried to the Hellenic base in the Saronic gulf, —shows that there was a certain amount of elaboration in the organization on the Greek side.
STRATEGY AND TACTICS.
When it is borne in mind that the Greeks were in this war carrying on operations on a scale infinitely exceeding anything of which they can have had experience, it must be admitted as remarkable that, whatever the defects of their military policy, whatever the mistakes they made, they managed to evolve out of their experience of operations on a small scale a system of organization which was applicable to the great operations of this war, and which did not in any known instance lead to a breakdown which can be attributed to defect in the system itself.
Of the strategy and tactics of the war itself little can be said which has not been already said in previous chapters; still, it may be convenient to collect together the considerations on this question which are suggested by the history of the two years’ campaigning.
Before doing so it is perhaps necessary to define as clearly as is possible the distinction between strategy and tactics, and that not merely in respect to the application of the terms themselves, but also to a not unimportant difference which may exist between the ways in which they may be employed.
Strategy is a term usually applied to the larger operations of war prior to or intervening between such times as armies are in close contact with one another.
When close contact takes place, and battle is immediately imminent or in progress, the name tactics is applied to the operations which then ensue, and which are necessarily on a smaller scale than those to which the term strategy is applicable.
A further and important distinction is this:—the operations known as tactics are the outcome of full consciousness on the part of the commander or, it may be, of the trained soldiers who employ them,— that is to say, that he or they know not merely what to do, but why to do it. The knowledge acted upon is in a sense scientific, though the reasons for the act may be various:—either that the operation has proved effective in the past, or that the tactics of the enemy demand it, or that it is called for by the nature of the ground. It is in respect to the last reason that tactics and strategy are most nearly comparable; and it is manifest that it is more easy to draw conclusions as to the results of action on a small area of country, all of which may be comprehended in one view, than in a large area such as would form the theatre of strategical operations.
The operations of tactics are so much more mechanical than those of strategy that it is difficult to conceive of them as being unconsciously carried out.
With strategy that is not the case. It may be conscious or unconscious. The strategical conditions of a country such as Greece, or indeed of any country, must ever be fundamentally the same, though liable to modification by the introduction of some great novelty into the act of war, such as long-range missiles. An army operating in Greece or elsewhere may fulfil the strategical conditions of the country consciously or unconsciously. It fulfils them
consciously if it appreciates them,—that is to say, if its movements are determined by them; but it may fulfil them unconsciously even if its movements are determined by considerations which cannot be said to rest on any strategical basis.
It is a very important question in the war of 480–479 whether either or both of the two sides operated with a conscious or unconscious strategy. Beyond question many of their operations were strategically correct. This every one who is acquainted with the story of the war must admit; but many of the greatest writers of Greek history have silently or expressly assumed that the agreement between the operations and the strategical conditions of the theatre of war was due to the circumstance that, in a country of such pronounced characteristics as those of Greece, the physical conditions were so marked that the strategical conditions might be fulfilled from motives wholly unconnected with them. This might undoubtedly be the case; but was it so?
Two great strategic designs are apparent in the main plan of the invaders:
(1) To create a diversion in the Western Mediterranean by stirring up Carthage to attack the Sicilian Greeks, and so prevent aid from that quarter reaching the mother country;
(2) To make fleet and army act in co-operation, and, furthermore, by means of the fleet, to maintain the command of the Ægean and its sea-ways.
On the side of the defence no plans of such magnitude are apparent; for the very good reason that the preparations to meet the invader were made too late for the carrying out of any far-reaching design, even had so composite a resisting force admitted of anything of the kind.
THESSALY AND THERMOPYLÆ.
The first strategic action of the Greeks was the expedition to North Thessaly. That was undoubtedly conscious strategy. It was undertaken under the supposition that the Vale of Tempe was the only entrance into the country; and had this supposition been correct
the strategy was sound enough. The mistake made was topographical, not strategical.
It was again sound strategy which made the Greeks, when they discovered their error, retire from a position which was indefensible with the numbers present with them. Even had their available force been much larger than it actually was, it would have been a mistake to try to defend a mountain line traversed by several passages. A successful attempt of that nature is rare in history, because the assailant is able to choose the passage on which his main attack shall be directed, whereas the defender has to distribute his force at all the possible points of assault. Moreover, the passage once forced, the assailant can generally threaten the lines of communication of the bodies of troops defending the other passes.
Thermopylæ and Artemisium display most clearly the strategy of the two contending sides. It did not perhaps demand much knowledge or much intelligence to fix upon Thermopylæ as a point of defence. It would be difficult to mention any other great highway in the world on which a defensive position so strong by nature is afforded. It was, on the other hand, a great strategic blunder on the part of Sparta that, having sent a small force to the pass, she did not later forward reinforcements. That was no doubt part of a still greater strategic blunder,—the desire to concentrate the defence at the Isthmus. The three days’ fighting at Thermopylæ proved conclusively that, had an adequate force been present there, the army of Xerxes would in all probability never have carried the pass. Sparta and the Peloponnese generally neither did, nor wished to, appreciate the immense strength of the position.
Did the Persians make a mistake in their assault on the pass? It would seem not. They did not at first know of any way of turning it. They had to take it, because it was the only route by which they could get their baggage train past Mount Œta. That is the difficulty of Greek campaigning. There are perhaps many routes in the country along which a light marching force in moderate numbers can make its way; but those by which troops with the ordinary paraphernalia of an army, and in numbers too large to live on the country, can go, are very few.
The Greeks learnt a lesson from their homeland,—a lesson which Xenophon expressed in striking language to the ten thousand Greeks when they began their famous retreat: “My view is that we should burn our waggons, that our baggage train regulate not our march, but we go by whatever way be expedient for the army.”231 In the present instance Xerxes was confined to the one route: he had to assault the pass, because in the then state of his knowledge of the country he could do nothing else. That he or his generals appreciated the importance of the path of the Anopæa so soon as they were informed of it is shown by the employment of the best troops in the army for its seizure and use.
It is, however, fairly questionable whether the direct assault on the pass formed part of the original Persian design at Thermopylæ. The delay of several days which took place before the actual assault was delivered, taken in connection with the contemporary events at Artemisium, suggests that it was originally intended to wait until the forcing of the Euripus by the fleet made it impossible for the Greeks to maintain their position in the pass for fear of the landing of a force in their rear If such were the design, it was one which bears testimony to the strategic capacity of him who conceived it. It was brought to nought by a factor outside all human calculation, the storm which broke upon the fleet when anchored at the Sepiad strand.
It was indeed at this period of the war that both the contesting parties gave conspicuous proof of their appreciation of the strategic conditions of the region in which their operations were being carried on. In point of conception there was little to choose between the plans of the assailant and that of the defender, but in point of execution the Greek plan was wrecked by the deliberate failure of those who were responsible for the land section of the general defence. That of the Persian was upset by causes beyond human control. It is noticeable that the invaders seem to have arrived at Thermopylæ with a fair working knowledge of the region of the Malian gulf and the North Euripus, and it must be suspected that the heralds sent to
GENIUS AND EXPERIENCE.
Greece to demand earth and water acted when there in a twofold capacity.
Whatever failure overtook the Greek defence at this time of the war was due not to the plan, but to the way in which it was carried out. The design to hold the land force of the enemy at Thermopylæ and his fleet at Artemisium was excellently conceived. It was almost certainly the work of Themistocles. Whether honest or dishonest, he was gifted with that rare genius which enables a man to take in the necessities of a situation vast beyond anything within his experience. The other Greek commanders were men of an ordinary kind, who would in all probability have frittered away the defence in a series of measures such as their limited experience dictated, or, still more probably, have concentrated at the Isthmus, where the strength of the land position would have been more than negatived by the weakness of the position at sea.
Eurybiades, the commander-in-chief, is unfortunate in the setting in which he appears in the history of the time. His must have been an unenviable and difficult part. The diplomacy he had to employ in order to accommodate contending policies would seem to both sides a proof of vacillation. Herodotus’ picture of him is not wholly sympathetic; and yet it makes it quite clear that he was intelligent enough to appreciate the genius of the man who was technically his subordinate, and diplomatic enough to give the advocates of the Northern policy a victory without provoking a fatal outbreak among the advocates of a different design. Nor is this a small tribute to the man’s capacity. The Peloponnesians, and especially the Corinthian opponents of the war-policy of Themistocles, were not people who could be kept in order even by the strong hand of Sparta, unless history draws a very misleading picture of the circumstances within the Peloponnesian league; and the man who captained the Greek defence through the troubles within and without of the year 480 cannot have been lacking in ability, even though he had at his side a pilot of the genius of Themistocles.
The failure to support Leonidas at Thermopylæ utterly changed the strategical conditions on which the design of the defence had been hitherto founded, and the practical surrender of the other
defensive lines north of the Isthmus completed the wreck of the plan. The Greek fleet on its return to Salamis found that no effort had been made, and no real intention had existed, to send an army even into Bœotia. It is noteworthy, however, that if the fleet really did suppose that the army was in Bœotia, and was not undeceived on this point until it arrived at Salamis, its commanders made a great mistake in not holding the narrows at Chalkis, and thus preventing the landing of Persian troops in rear of the most eminently defensive passage in Bœotia, that long strip of narrow land between Helicon and Kopais which extended from Chæroneia to Haliartos.
The strategy which was forced upon Themistocles by the state of things which he discovered in existence on his return to the Saronic gulf was the strategy of despair. The position taken up at Salamis could only be justified on the plea that there were no other narrow waters between it and the Isthmus where the Greek fleet could be sheltered from the disadvantage of meeting a more numerous fleet in the open, a large portion of which, the Phœnician, was probably superior to it in manœuvring power. Furthermore, if the Persian commanders took the view that the fleet at Salamis must be defeated before an advance to the Isthmus were attempted, and detained their ships on the Attic coast, then the Persian land army, unaccompanied by the fleet, would be rendered incapable of any sustained attack on the fortifications which the Peloponnesians had erected. But that “if” was one of terrible significance, and the evidently nervous desire of Themistocles to bring about a battle must have been due to his recognition of the precarious position in which the Greeks would be placed supposing that the Persian fleet did choose to ignore them at Salamis. The most fatal weakness in an altogether dangerous situation was that a large portion of the population of Attica was on Salamis Island, and could not possibly be left to the mercy of the foe. How it came about that the Athenians gave such hostages to fortune, instead of removing them to the coast of Argolis, can only be conjectured. The probability is that the interval between the arrival of the Greek fleet at Salamis and that of the Persian at Phaleron had been too brief to render it possible for the whole Attic population to be transported across the Saronic gulf.
SALAMIS.
The blunder which led Xerxes to attack the Greeks at Salamis was fatal alike tactically and strategically. He had the game in his own hand, if he could only have recognized the fact; but in his confidence of success with the forces at his disposal, he wished not merely to out-manœuvre but to capture the whole Greek fleet.
The results of Salamis were immediate. The defeat and moral disorganization of the Persian fleet made it incapable of maintaining its position on the west side of the Ægean, though in point of material damage, relative to numbers, it is probable that it had not suffered more severely than the fleet which had been opposed to it. Its departure withdrew, as it were, the keystone of the Persian plan of invasion; and the whole edifice of design fell into ruin which was incapable of repair, though the wreck was not so complete as to render it impossible for Mardonius to make use of the materials in the ensuing year. The blow had fallen on the indispensable half of the invading force; and, bereft of the aid of the fleet, the land army could no longer maintain itself in a country whose natural resources were wholly inadequate to supply its wants.
According to the account followed by Herodotus, Themistocles proposed that the Greek fleet should immediately take the offensive on the Asian coast. The evident confusion of the story as to date renders it difficult to say whether the proposal was a wise one or not. It was, however, rejected by Eurybiades, who up to that time had displayed sound common sense; and if it was really made at the date to which Herodotus attributes it, there can be little doubt that Eurybiades was right. The time for offensive operations had not come; for anything of the nature of a reverse on the Asian coast might have restored once more all the evils of the previous situation in Greece; and it is quite possible that some argument to this effect put forward by Eurybiades has been translated into the form of his answer to Themistocles as given by Herodotus.
The actual design which Mardonius had in his mind when he persuaded Xerxes to leave with him in Thessaly the most effective part of the army of invasion is beyond conjecture at the present day. Possibly he had at the time no very definite plan, but was content to guide, and be guided by, events. His retirement to Thessaly was
certainly due to the question of commissariat. The country was infinitely richer than the rest of Greece, and, besides, he had to organize a new line of communications along the North Ægean coast. With this intent, probably, Artabazos accompanied the retreat of Xerxes as far as the Hellespont. The winter gave Mardonius time to form his plans for the next campaign, and though Herodotus does not profess to know what the plans were, the operations during the summer of 479 give something more than a clue as to their nature.
Mardonius seems to have formed a design of two alternatives.
He knew, on the one hand, that unless the Persian fleet could be brought back across the Ægean, an attack on the Isthmus would be hopeless, if not impossible; in other words, the complete subjugation of Greece was out of the question. But so long as the Greek fleet remained as powerful as at that moment, it was also hopeless to expect the return of the Persian; and the first problem he had to solve was how to rob the Greek fleet of its strength.
With this intent, negotiations were opened with Athens, through the medium of Alexander the Macedonian. When this line of action proved ineffective, he brought pressure to bear upon the Athenians by once more invading their territory and occupying Athens. This again had no effect, except to make them even more stubborn in their refusal to treat.
DESIGNS OF MARDONIUS.
The first alternative plan had failed at its outset, so he now resorted to the second.
After burning Athens and committing devastation in Attica he retired into Bœotia, using Thebes as his base of operations.
There are two remarkable facts in connection with this movement
The disaster at Salamis, and the consequent retirement of the Persian army to Thessaly, had not shaken the loyalty of Bœotia to the Persian cause, and this despite the fact that the Bœotians were not at the end of 480 deeply committed either to the Persian, or against the Greek. Bœotians had, indeed, been present in Xerxes’ army when it invaded Attica; but there is nothing whatever on record which can favour the supposition that the Bœotians had actually
