Bits are Free Bytes are Not: The Dawn of a Regulated Flexible Spectrum Access (Draft v0.99.02) Maximillian Sandmann, R V Prasad, Peter Anker, Ignas Niemegeers
I. Introduction The idea of cognitive radio (CR) as an alternative wireless communication paradigm, capable of managing and executing itself in real-time without human intervention, was first proposed by Joseph Mitola III and Gerald Q. Maguire, Jr. [XXX]. In the utopian world of ubiquitous computing, each device would be equipped only with CR, thus empowering it to communicate on any available spectrum. However, the question arises where to look for the unused spectrum with the myriads of wireless devices employed around the globe and with a status quo in the spectrum regulations. As demand for radio resources increases, industry and academia are searching for efficient methods to share the spectrum â€” the wireless domain. Historically, spectrum is regulated through the assignment of exclusive rights. Spectrum is split into fixed blocks, which are designated to a specific service or technology. There are bands designated for, e.g. analogue radio broadcasting, air traffic control, emergency services, television broadcasting, and mobile radio. The frequencies are exclusively assigned to licensees to offer these services. This model, which was introduced in 1927, is still the major model followed in the current wireless world, but it is inefficient. This observation was strengthened by a study of the National Science Foundation with the cooperation of the Shared Spectrum Company to find the extent to which the spectrum is currently used [XXX SS]. Measurements in the 30â€“3000 MHz band show that utilization of some of the radio channels is less than one percent, whereas the average occupancy over all the frequency bands is only 5.2 percent. In fact, except in one case TV37-51 (608-698MHz) none of the other bands considered by SSC [SCC Web page] averaged over the six locations crossed 25 percent. Surprisingly, the maximum total spectrum occupancy for New York City during one such measurement was found to be only 13.1 percent. Even during the peak hours of usage, one could find free spaces in the spectrum of the public bands. Thus, we may conclude although dividing spectrum into exclusively assigned blocks creates certainty for service providers it is inefficient and the rigid allocation policies of regulators have resulted in an artificial scarcity of radio resources [XXX Vps]. From a regulatory perspective, there are two different models considered to govern the spectrum: a model based on exclusive rights, as we described above and a model based 1
on open access (Faulhaber, 2006). In the first model, the rights to use the spectrum are defined with as few usage restrictions as possible. This model has its merits for operators which use spectrum to sell a service, like mobile telephony. However, this model is not particularly suitable for end-user deployed applications like remote control or wireless local area networks. The second model is most of the time loosely denoted as a public spectrum commons. In a public spectrum commons anybody can have access to spectrum without the need for a license. There are only general conditions imposed to keep the interference of the other users at an acceptable level. The spectrum commons may be open to all or restricted to a specific group of users. This is a way to ask the users to organize themselves. While there are many success stories of unlicensed access like the 802.11 standard, there too are some aspects that may prove counterproductive. Although the standard is accompanied by some etiquette and protocols, there is uncertainty about the amount of users that can be accommodated. What will be the efficiency once these bands start becoming more and more occupied? A simple analysis could infer that the efficiency of such a system also depends on the density of the nodes sharing that band. Further uncontrolled unlicensed resource sharing could lead to the phenomenon called “Tragedy of Commons” [Hardin, 1968] once the bands it operates in become saturated. Advocates of the public spectrum commons expect that the tragedy of the commons can be overcome by smart radios with embedded protocols to manage their frequency use without interfering with others. This can be realized, with a cognitive radio which is aware of its radio environment and only accesses the unused spectrum, or in an “opportunistic” way. In fact, models for this opportunistic spectrum access (OSA) are probably as old as radio communication itself. One of the first communication systems using shared radio resources was maritime communication (1910s). A detailed study of the history of the shared spectrum usage is given in [xxxx]. Current ideas about Opportunistic Spectrum Access (OSA) are based on the existence of white spaces, i.e. frequencies assigned to a primary user, but, at a particular time and in a specific geographic location, not used by that primary user. The objective is to utilize these white spaces without interfering with the primary user. Once the spectrum holes are chosen by more and more users, the problem returns. Beside there is a catch. There is no long-term certainty in the access of spectrum. The radio might come into a situation where it can not find a free chunk of spectrum for its communications. This will limit the quality of service that can be offered in a spectrum commons. As for both models, exclusive right and public commons, it can be concluded that there is a need for efficient spectrum management recent proposals concerning the usage models of cognitive radio also include Dynamic Spectrum Access (DSA). Dynamic Spectrum Access can be basically defined as a near-real-time adjustment of 2
spectrum usage in response to changing circumstances. The core idea is that by utilizing spectrum in a dynamic manner, utilization can be greatly improved. However we note here that, in case of the exclusive rights model, the “Tragedy of Anticommons” could still occur. [Science, Heller]. Heller states that exclusive usage of a scarce resource may lead to a situation where there are too many holders of exclusive rights. The rights become too fragmented and an effective usage of the scarce resource in not possible. Thus both of the proposed models have their advantages but there also some issues. The question now is: is there a way to combine these models? Further we can't ignore the fact that users would like to have some assurance – everything cannot be left to the selforganizing capability of users. Further there's a continued discussion about the real value of spectrum and how rights should be distributed. To quote Viviane Reding, European commissioner for information society and media [Joint dinner of the European Regulators Group (ERG) and the Radio Spectrum Policy Group (RSPG) Gothenburg, 27 February 2008]: “Most ‘valuable’ does not mean only the most ‘profitable’ services. We need to think in terms of optimisation of spectrum in a wider sense, integrating social, cultural and economic aspects. As the European Commission also stressed in our recent Communication on the Digital Dividend, there is a necessity to shift our focus from technical spectrum efficiency to an optimisation in terms of the value to society of the services underpinned by the spectrum.” She reflects the EU’s policy change “more competition, more services and more choices” but explicitly adding on the basis of public interest”. The FCC also talks of “in public interest”. This is a significant aspect to note. While the regulators or owners hitherto, are ready to make drastic policy changes they also like to be guarded in their approaches. The need for this discussion also stems from the fact that the Dutch government has published a spectrum memorandum in which on the one hand a license is regarded as a property right which gives more freedom to operators to offer the services they find suitable and on the other hand more room is given for spectrum access without the need for a license. The interesting situation arises that the Netherlands Antilles intends to open up parts of the TV band that are not used for TV itself, the so called TV White Spaces for other uses [NA-Website]. Thus it is a very opportune time (see also Ofcom [http://www.ofcom.org.uk/consult/condocs/cognitive/statement/statement.pdf]) for a new vision on efficient spectrum use in the future. This article intends to initiate a discussion as to what could be done and it takes the specific example of TV Bands and their usage as a futuristic spectrum usage model.
ROADMAP Before we discuss our proposal, we want to introduce first the issues and models in allocating spectrum access.
Some Important Terms and Definitions according to IEEE P1900 Since we are talking of radio spectrum access, we should have the background in what has been recently proposed. We should take a look at the multiple types of spectrum access that have been proposed. We mention here some of those terms with the standardized definitions [P1900] to avoid the confusions in the terminology later in this article. Dynamic spectrum access [P1900]: The near-real-time adjustment of spectrum resource usage in response to changing circumstances and objectives; including interference experienced or created; changes of the radio state (operational mode, battery life, location, etc.); and changes in environmental/external constraints (spectrum, propagation, operational policies, etc.). Opportunistic spectrum access [P1900]: The method by which spectrum users operating on a secondary (and possibly unlicensed) basis within a frequency band with designated primary (and possibly licensed) users exploit unused in-band segments for their own purposes without causing interference to the active interference-intolerant primary users for the duration of the availability of the spectrum in question. Software-defined radio (SDR) [P1900]: A type of radio in which some or all of the physical layer functions are software defined. For example, it is feasible that a radio has two or more modulation schemes, each of which is implemented in hardware. However, software control is used to switch the radio characteristics. In the FCC and ITU-R definitions, if such control is used in a radio, it is considered to be SDR, even if the radio signal processing is done entirely in hardware. IEEE 1900.1 does not consider software control to be SDR. Cognitive radio: [References] Cognitive radio network [P1900]: A network capable of establishing links between its cognitive radio (CR) nodes to establish connectivity and to adjust its connectivity to adapt to changes in environment, topology, operating conditions, or user requirements. White Spaces: Unused electromagnetic spectrum, for example, between TV channels. Originally set up as protection bands to eliminate interference, wireless microphones used in theaters and other venues have employed this spectrum. [See White Spaces Coalition.]
II. Cognitive Radio â€“ Possible Future Usage Models A. Coase's Theorem The market model we mentioned in the introduction is based on the assumption that spectrum is a commodity used by, e.g. a service provider to provide services. Depending on its subscription demand the service provider can choose to buy or sell spectrum usage 4
rights. This model is based on the work of Ronald Coase (Coase, 1959). He posed that the allocation of the spectrum should be determined by the forces of the market rather than as a result of government decisions. Radio licenses should be seen like any other scarce resource in our economy, such as land or labor. Thus the market should not only decide who will have the licence but also what services should be provided and if a business model fails the rights can be bought by another operator with a different, more successful, model or by a new entrant on the market. Here we refer to [ETSI MARTIN CAVE NEW BOOK CHAP] which consolidates Coase's idea that “if property rights are well defined and understood then users (operators and their customers together) will be able to negotiate rights amongst them and use the resource efficiently” [Coase, 1960]. So in order to make this Coase-model work successful, its transactions should be efficient, and transparent ownership information is necessary for market participants to minimize transaction costs. At the moment already a lot has been done on the regulatory front to realize more flexible and efficient use of the spectrum. On a worldwide level the ITU-R carries out studies to enhance the international regulatory framework to cater for more flexibility in the use of spectrum. The ITU-R also carries out studies to see if there are any regulatory measures needed to enable the introduction of software defined radio and cognitive radio. In both the U.S. and in Europe, changes are made to the spectrum regulations to allow for more flexible use of the spectrum and to allow spectrum trading [Anker, tbp] B. TV White Spaces The other model is what we can loosely define as “opportunistic spectrum commons” where the opportunistic acquired spectrum can be shared by all the competing parties. In this license-exempt system, white space access is allowed under some restrictions. The FCC has some guidelines for this case which we discuss in the next subsection. In this case cognitive radios are allowed to make use of any unused spectrum it finds as long as it follows the guidelines of the FCC. Strict rules, etiquette, and conformance of the CR devices with the standards are needed to realize this opportunistic use of the spectrum. On November 4, 2008, the FCC adopted a Second Report and Order and Memorandum Opinion and Order on the matter of unlicensed usage in TV broadcasting bands. This allowed unlicensed radio transmitters to operate in the television broadcast spectrum at locations where that spectrum is not being used by licensed services. This unused TV spectrum is often referred to as “white spaces1. It is shown that at many geographic locations across the U.S.A, an amount of “white spaces” in TV channels 20-51 is available [showmywhitespace.com]. The TV White Space (TVWS) devices have to adhere to a strict set of rules. 1
FCC 08-260, Second Report and Order and Memorandum Opinion and Order, Para 2.
Some noteworthy rules of the use of these bands are: 1) Fixed devices are allowed 1W transmission power and 4W EIRP while Personal or portable devices are allowed up to 100 mW EIRP; 2) TV channels 2, 5-36, 38-51 can be used nationwide for fixed devices. TV channels 21-36, 38-51 can be used nationwide for portable devices; 3) Devices must sense signals from licensed primary users up to -114 dBm; 4) Daily consultation of a database which contains all licensed use in the TV bands by the fixed devices or when the location is changed. However at the moment there are no rules to limit the interference between the TVWS devices. Further, it should be able to accept the interference from other FCC approved devices. The assessment is that TVWS would also see a large amount of devices just as WiFi. It is argued that due to this unlicensed usage, TVWS might not achieve the desired or maximum amount of spectrum efficiency in every geographic area. Further the decreasing amount of clean channels due to the implementation of more broadcasting or IMT services [ITU-R Report M.2078] accompanied by the presumably ever-increasing demand of bandwidth by the end-user in these channels can lead to the conclusion that already in many populated areas today it is not possible to exploit the true potential of unlicensed TVWS networks unless there are some mechanisms. Thus to make spectrum usage in TVWS more effective we may have to propose a different methodology for TVWS use.
III. Issues Before we go into the discussion of our proposal, let us talk about the viewpoint of the general public and the governments. One could argue that with the introduction of spectrum commons also came the viewpoint that spectrum under certain conditions should be free for usage. While there are multiple possibilities for the usage of the spectrum and services offered over the spectrum for which one should pay, the spectrum can also be thought of as a natural resource like air that is everywhere available to everybody. On the other hand, most governments, hitherto, are of the opinion that it is their property and prerogative to use or sell the spectrum. This fact is also due to the evolution of communication and applications using the spectrum which evolved during the world wars. The prerogative of using it for national security also mattered, it seems, for governments to take this view. However many studies have concluded increased spectrum usage benefits the overall economy in both property rights [Fuss, Meschi, Waverman, 2005] and commons models, one could therefore argue that the easiest way to increase spectrum usage, is to reduce its cost to the overall public. In fact a recent study commissioned by Microsoft [The economic value generated by 6
current and future allocations of unlicensed spectrum, Perspective Associates] estimates the economic value generated by WiFi in US households, currently between 4.3 and 12.6 billion dollars annually, could increase by opening up the TV bands for WiFi usage (i.e. free of spectrum fees) with an additional amount of up to 14.9 billion dollars. It is therefore no surprise that governments increasingly are starting to think about liberalizing their hold or policies regarding their spectrum [WRC 2011 Agenda Item 1.2]. Notwithstanding one must also note with respect to liberalization, that with ‘own and use’ policy many operators may feel the need of achieving a gain over their initial “spectrum investment” for their shareholders and as a result might overprice this resource to their customers. Thus the economic as well as social benefit, which we already discussed, can be reduced. A further issue is that although, many of the projects addressing the problem of spectrum scarcity are using the cognitive approach to use the spectrum efficiently, omitting all control and allowing the users to manage themselves may only work in small measures. Once the spectrum holes are selected by multiple users, the problem can persist as how they should be shared. Opening up the entire spectrum for unlicensed use, as an extreme case, may be plausible with the improvement in technology; however, we have seen that it may not work all the time. Thus we have to think of a different model and an essential aspect in the new policy should be to achieve a greater benefit to the overall economy as a result of this important resource.
IV. Bits are Free Bytes are not2 Our Philosophy With the earlier mentioned discussions on the economics of the spectrum, issues and requirements, we come to a point as to how to fulfill acceptably close expectations of everyone. Further, we also need to take into account the public demand and usage of the spectrum. Any civilized society will strive for a constant improvement of the quality of life. The network of roads built by the government is free for use by the pedestrians or cyclists. However, if a person uses a car on those roads, he would be paying some cost for it. The idea there is arguably to support the small consumer but those who would like to have the resources for higher levels of usage need to pay for it. Now, future communication networks, in particular the Internet, are becoming important basic facilities for the public. It will therefore not be the exclusive forte for those who can afford to use it. If the standard of living has to be improved then the basic communication needs of the public should be supported and to a certain extent be free. 2
First quoted in Nico Baken, Nico van Belleghem, Edgar van Boven and Annemieke de Korte, “Unravelling 21st Century Riddles – Universal Network Visions from a Human Perspective”, The Journal of The Communications Network, Volume 5, Part 4, October–December 2006.
Although in recent times many proposals with this principle in mind have been done by proponents of spectrum commons models [White Space Coalition], we believe that this principle could be also applied to the exclusive rights model. In the next sections we will therefore propose a system which should stimulate to access lower levels of bandwidth, or bits for free, although we still assume that for higher levels of spectrum usage (e.g. watching a movie) a price has to be paid. Instead of individuals, operators may also be allowed to use this spectrum without charge, and for the sake of providing free access to the customers and the citizen to support basic communication services such as Internet, email, SMS, etc. The offering of services can be made more dynamic by the offering of a mix of free services and premium services which are paid for. Based on operatorsâ€™ business models, some scenarios that we can visualize at this point of time are, (a) advertisements, personalized, location based or both in lieu of spectrum access at cheaper or no price; (b) femtocells, where the operational cost can be largely transferred to the end-user or (c) a combination of both. The scenarios above conclude that there are possibilities for service providers to gain a net benefit over their direct operational cost accompanied by these offerings. However in the current system where operators have to participate in auctions to acquire their spectrum, operators aren't given an incentive by the regulator to do so. In fact as spectrum currently goes to the highest bidder one could argue the exact opposite. Thus an essential part of our proposal will be to come up with a system which adequately does so. A further objective is that the implementation of cognitive radio should enable us to design a system where commons and exclusive rights are able to better coexist among each other.
V. Guiding Principles for the System design A. Guiding Principles Now, the discussion is all about how to go about designing a system that could fulfill the needs of the consumer, operators and governments. The order in which it is mentioned here should also give some hints as to who should be given the priority -- the consumer or the operator? The design of such a value system should take into account the aspects which have been discussed in the previous section. We shall briefly touch upon the study by Martin Cave and others in [spectrum regulation Martin Cave]. The following are the principles sieved from [spectrum regulation Martin Cave] with regard to spectrum usage: 1) Ensuring efficient use of spectrum 8
2) Creating incentives for investment and innovation 3) Policies encouraging competition 4) Non-discrimination 5) Transparency 6) Workability 7) Providing long-term certainty 8) Minimum interference 9) Ensuring compatibility 10) Public interest 11) Recovering administrative cost 12) Adhering to international regulatory agreements [martin cave] While we have the above guiding principles which will provide the system with some checks and bounds, we now have to gear up for designing a system that has the characteristics of liberalization. The governments, many in the West, have a reasonably liberalized market economy and reduced their hold on the natural resources and the nature of the transactions between people and entities. However once the spectrum becomes a public resource, addressing the conflicts in sharing them is a question. As we know that the spectrum as a resource can become scarce and thus adds a value to the one who owns it. The basic idea should be to build a system that avoids conflicts rather than resolving the conflicts later. The premise is to build the system, conflict free. The question is how to build a system which can assure fairness to the parties involved in the transactions. Of course the transactions should be competitive as mentioned earlier. Finally, in the context of spectrum usage and management, the interference should be kept at an acceptable level. We use these guidelines in proposing our system later in the paper. B. Value Chain For an operator, spectrum is a means to do business. Any business has a value chain, which is the relation between the continuum of cost incurred and the revenue generated at the end [Keith][ Michael Porter, Competitive Advantage: Creating and Sustaining Superior Performance.]. A value chain is a chain of activities. When the products go through many stages of activities, they gain value. This final value is higher than the sum total of the added value at each stage. The cost includes everything â€“ from the investment in raw materials to operational cost and maintenance cost. The Cognitive Radio Network (CRN) and in general cognitive radio can also be seen as part of a value chain since it allows assignment of value to the spectrum and moreover the value could be modified by devising different methodologies. In fact, our proposal addresses the dynamic modification of the value chain through the use of cognitive radio. Generally, the value of a good, commodity or service is ultimately subjectively â€˜valuedâ€™ by the consumer. It mainly depends on the supply and demand. The trading that happens between the consumer and the seller or service provider is essentially a system that enables negotiation between them to fix the value. An important aspect of any trading is a certain degree of volatility so that pricing is able to reflect actual market conditions. And 9
since SDR and CR technology enables a higher rate of change in its operation, the idea here is to allow the consumer and operator who use the spectrum to have a level playing ground to constantly and dynamically value the spectrum. The freedom of the customers to select an operator, coupled with operators who are competitive, with the least amount of investment, should lead to a useful and efficient use of the spectrum. The value of a system or a technology can be enhanced if it is dynamic. An analysis of the value change for adaptive systems compared to non-adaptive systems has been done in [Keith]. It is also necessary that competition is allowed and checks and bounds on the usage of resources are present. Though in many circumstances the cost of the resources cannot be easily quantified, the dynamic usage will help in allowing the system to get its real evaluation. This motivates the design of our system explained in following sections.
VI. Proposed Methodology INTRO A. Three Level System Model
Figure 1: The System The system can be broadly seen as having three levels. First is the regulator and real-time allocation system. The second level consists of the Operators. Operators interface with the customers and define prices to the customersâ€™ requirements. Operators provide the required support to all their customers in return of the costs imposed. The third level is the Customers (also referred to as users), who may have SDR capability in their devices. The allocation system is managed independently and is accessible to all the operators. The system is transparent in its decision making. The regulator may change from time to time the algorithm, and its inputs such as power constraints, spectrum bands for the use and other settings of the allocation system. 10
As is shown in Figure 2 one can see that the operators, after an initial assignment of spectrum, where they all have some limited working spectrum at the start, may change their total bandwidth based on their requirements from time to time, e.g. operator O 1 has reduced his requirement while O2 and O3 have acquired more spectrum. The real-time allocation system should be able to support this change. We explain the allocation system in more detail later.
Figure 2: A representation of changing frequency occupancy with time by different operators The idea here is to have a regulatory body that works as an arbiter but still leaves it to the operators and their customers to manage themselves with a transparent system. The regulator has a loose hold on the allocation system as shown in Figure 1, in order to allow the enforcement of regulations and etiquette. Regulators being part of the government could also change the policies under circumstances such as emergencies, etc., which will be transparently accommodated into the allocation system and change the way the system works. This also helps in the case of disputes by allowing the regulator to be an arbitrator too. In normal circumstances, the regulator may be simply represented by a database of rules and guidelines for the allocation system. The regulator will have to act under a certain set of norms. An operator may work in different bands which are not contiguous. Changes in the usage of spectrum by the operator can happen with a minimum time bound, Tmin on T based on the technology, which may be set by the regulator. In fact, if we make the parameter T very large (say years) after the first assignment (through allocation), it becomes slightly similar to the model currently used where operators acquire rights â€“ whether they use it or not. Therefore, similar to a minimum bound on T, there would also be a maximum bound Tmax on T, which is also dependent on technology and the usage characteristics of customers. Since, the system is transparent and the participating devices can be 11
registered, the changes made thereof to the system should also be transparent and negotiated with the operators. A 2
Figure 3: Spatial reuse and allotment of spectrum We can replicate this methodology easily in different geographic domains, let us call them cells. The allocation system would be communicating with the adjacent cells, in order to reduce interference. Therefore he spectrum used in the adjacent cells could be reused elsewhere. This representation is shown in Figure 3 for four adjacent areas. Thus a cooperative spatial reuse of the spectrum is also dealt with. Now the question is where is the opportunistic access? The spectrum that it is not allocated to one of the operators could be utilized by the opportunistic users. They donâ€™t pay for it but would use the spectrum under strict regulations and vacate the system as and when that part of the spectrum is allocated to an operator. B. Assumptions INTRO There are some assumptions we have made here: 1) The frequency allotment to an operator is contiguous for the sake of making the system simple to explain. Thus we have shown it growing and shrinking in only one direction. However, this need not be rigid, for example, the third row in Fig.2 shows that the Operator 2 and 3 have shifted their usual allocated spectrum in both directions and are in each othersâ€™ domain compared to the first two rows. 2) The operators will not pay for the spectrum they get through the regulator, but instead will pay fees based on the transactions with their customers. This model, 12
though not directly proposed here, is a method by which operators are discouraged to hold more spectrum than they could use. 3) A fee on spectrum usage is the way the regulators could make up for the operational costs. This follows from the fact that the spectrum is not owned by anyone but for a fair and efficient use there must be some checks and bounds and there should be a custodian. This system may cause less interference compared to the usual OSA systems where in some cases is done by the secondary systems. The information on the spectrum allocated to the operators is available publicly, everyone can get the status of the parts of the spectrum that are in use -- with rights as well as the ones that are claimed opportunistically. This helps as a warning system to the secondary usage whenever the spectrum is allocated. The secondary systems still need to vacate on the arrival of a primary user or when the spectrum is allocated to another party by the allocation system. The byproduct of this system is that sensing becomes easier since a centralized repository knows what has been allocated and for how long. To this end the detection complexity would be reduced. When it comes to the spatial usage of the spectrum, one can apply a similar methodology and expand this model with space as another parameter.
VII. The Components A. Allocation System We envisage a distributed allocation system operated by the Regulator. However, we use a centralized and single system for explaining the concept. It is a system that enforces differential fees by considering, as discussed earlier, the various aspects of the regulator's guidelines. Thus in the creation of a market, the allocation system should take into account many aspects, such as, fairness in allocating the spectrum, the tax that needs to be paid to the regulator, the historical usage of the spectrum, the bandwidth requested, the time interval between two allocations â€“ a constant or variable, Tmin and Tmax as described earlier - period for which the spectrum is allocated to an operator, encouraging the efficient use of spectrum, etc. As an example, we may say that an operator who uses a directional antenna to increase the bit rates would be rewarded in some way when getting the spectrum allocated. It is a system that allocates spectrum to operators with financial transaction involved however it gives incentives for instance on the basis of taking only the amount of spectrum that is required. The incentive would then be distributed on basis of who meets that requirement the most. This system should be transparent and should cater to the requirement of spectrum allocation and reclamation in real-time. There are many methods with which one can maximize the gain for one of the players in the above system or a fair gain to all of them. However, the main criteria for the regulator are, to allow competition, reduce the start-up investment by the operators, encourage the efficient use of the spectrum by cutting down fees. 13
The operators would be allocated a minimal amount of spectrum in the beginning for them to operate. The remaining spectrum will be taken based on the requirements in realtime. This is called the second priority spectrum, i.e., the spectrum that is acquired through allocation will again lie with the regulatory authority when unallocated to fix the conditions for allocation to operators for use. B. Incentive Pricing We propose therefore that the regulator no longer allocates spectrum to operators on the basis of allocation it to the highest bidder, but instead solely on the choice of the consumer, or his transactions with operators. If an operator gets more transactions, it can acquire additional frequencies from the regulator. However, the regulator will levy a variable percentage based incentive â€œtaxâ€? when they make a successful transaction with their consumers. This tax paid serves three purposes. The main guiding principle for spectrum pricing is efficiency [Cave, 2007]. The regulator can use the tax to promote the use of more spectrum efficient technology. The use of more efficient technology will come with a cost and the tax will give an incentive to use more efficient technology, and will reflect the scarcity of the spectrum, i.e. the demand for spectrum against the amount of spectrum available. The incentive, if properly designed, could also be used to realise other social objectives. However, the regulator will have to be very careful in the use of a spectrum management tool for the realisation of other objectives. As we will discuss in the next paragraph as well, the tax can be used to recover the costs of spectrum management by the regulator since spectrum management by the regulator will come with a cost. As discussed earlier this spectrum allocation, with a maximum lease of Tmax will guarantee that a certain volume of trade will be realized. This will generate price signals on a regular basis as a result of which price will better reflect the underlying value and can consequently lead to a better functioning market. Putting the consumer's decision making ability central should lead to a better market overall and at the same time enables the regulator to achieve all the objectives we discussed in the guiding principles section.
VIII. The players A. Regulator/Government In our model both regulator and government are seen as the same entity. We use regulator to mean both. The regulator is the custodian, not the owner of the spectrum. The regulator 14
should have a couple of advantages compared to their current model because of our system. The following are some of the important objectives regulators could achieve in our proposed system: 1) Potentially maximizing the use of spectrum it operates in, because all the participants are able to use each others empty spectrum. This could be applied both to the already allocated (e.g. IMT service bands) and currently unused bands. 2) Opening up in a short time span a countryâ€™s entire spectrum for usage. 3) Encouraging higher competition among operators, which most likely leads to lower cost for the end user and specialisation at the operators. Thus more overall demand and economic benefits can be achieved. 4) Forcing the operators to broaden their scope and invest into areas where higher margins can be gained. This can lead to a lot of new products and has an overall positive economic effect. 5) Reducing the regulator's overhead of spectrum allocation, as after the initial assignment most responsibilities are handed over to the allocation management system. This process can lead to a lower cost for the end-user. 6) Encouraging the efficient use of spectrum as incentives give the regulator more power to achieve its targets concerning spectrum usage. Incentives may be the duration of use, guarantee of spectrum availability, the amount of taxation, etc. 7) Efficiently catering for emergency situations. This is achieved by the allocation system. In the affected areas and for the required duration a part of spectrum bands would be kept unassigned. This spectrum would be directly allocated to the first responders. Since the information about allocated bands, for emergency purposes, is available with the regulators (here in the allocation system) current OSA sensing techniques need not be used. This makes emergency response faster and interference free. Thus, the proposed system could in fact support emergency situations in a better way. As given in the section on guidelines, the regulator is paid through a tax on the usage of the spectrum to cover its overhead for maintenance and operating costs [ITU-R SM.2012]. It is believed [refer to the guiding principles Sec] that the regulators will eventually need to operate under this broader framework of allowing competition as well as making sure that the end customers get some benefit, while the regulators maintain and enforce the framework. B. Operators Because the allocation system is designed for CR and SDR devices, thanks to the ease of acquiring spectrum and the related incentive pricing there are possibilities on the operators side for highly specialized service offerings, for instance offerings to tourist only.
For the operators this system also lowers the threshold for accessing the market place by eliminating the entry-license cost a “greenfield” operator normally faces. This is an important aspect in encouraging new services – while starting up a new network or an incumbent extending an old network. The system provides overall flexibility. The amount of allocated spectrum always reflects the current demand of an operators’ network. This also ensures that new operators would get into the system with fewer hurdles to increase the competitiveness and ultimately the customers will get the benefit [google competitiveness Blog]. For the incumbent operators the right value of a certain amount of spectrum is always being reflected. Since the biggest quantities of mobile radios in the world are designed to operate in IMT bands, these bands might be more valuable than others. This is attractive for operators which have already made investments in these bands. The system supports heterogeneous networks. Therefore incumbents are able to maintain their original network infrastructure. However they might face an increased competition. C. Customer For customers, the system provides an opportunity to use services purely on an actual demand basis rather than subscription basis. The system also provides OSA. In fact an entity knows when the spectrum is in use and when it is not. With the improvement in technology one can visualize the dissemination of the information regarding the available unused spectrum which could be used for some time. This system bestows information transparency, or makes it easier to find out which network is the most attractive. This system also allows the freedom to choose between operators (networks), the amount of bandwidth needed, and the time slot(s) over which the customer wants to purchase rights at a micro level. For example, a customer could take the voice call service from one operator and broadband access from a different operator based on various criteria. This is in fact our vision of customers that are free to purchase what they want and from whoever they want without any binding or subscription.We stated that the customers are connected to a specific operator and all the services are offered by a single operator according to Figure 1. However, we envisage that customers, in possession of a CR device could select in real-time services from an operator who gives best ‘value’ for the money at that instant. It is not difficult to see that customers themselves could in fact access a centralized repository, which broadcasts the price and constraints of a particular service. Then based on the information customers may go to a specific operator for acquiring the service.
IX. Way Ahead – Possible Implementation A. Proposed Model for Netherlands Antilles Since the UHF TV bands are to a large extent not used in the Dutch Antillean islands, it is on one hand it is easy to suggest a completely new model, however on the other hand it is also difficult to predict the outcome. We try to propose a model for adoption here. 16
We propose to: 1) Issue licenses for operators who will cover the whole geographic area (in this case island-wide); 2) Allocate spectrum to operators within geographic areas and approving transmission powers to cover those areas (as long as they are within safety limits); 3) Give every participating operator an equal amount of first priority spectrum; 4) Allocate an operator a secondary priority right to use other operator’s empty/unused spectrum, without causing interference to the other operator's existing operations; 5) Apply a variable taxation on the transactions done within the system; 6) Assign, in case of necessity, exclusive channels for unlicensed access only; 7) Allow consumers to switch between the networks of operators, in general, this is done when there are more favorable conditions in another network; Spectrum allocation rights can be divided in the space, frequency and time domains, e.g. time slots of or minutes, hours, days, weeks or maybe even months. B. Possible Implementation The work on CR and SDR has opened up a new avenue for dynamic and spatial aware radio. Thus a strict division of frequency of operation is not necessary anymore. Further, in the light of Netherlands Antilles opening up the TVWS for new usage models and, the Netherlands government’s policies on the introduction of more flexibility, it is the right time now for proposing and to a certain extent experimenting with new methods to share the spectrum.We envisage that the following technologies could be used for a limited demonstration of the ideas generated here. 1) WiMAX Analog TV bands (700 MHz) are envisaged to be used by WiMAX [http://en.wikipedia.org/wiki/WiMAX] once the roll-out of digital TV happens. There are positive indications for this, for example, the digital dividend concept [ XXX] by the ministry of economics of the Netherlands. EU commissioner Viviane Reding has suggested re-allocation of 500–800 MHz spectrum for wireless communication, including WiMAX. Ofcom director Kip Meek developed a plan for the 2G and 3G spectrum. The 2.6GHz spectrum for Wimax and LTE and the digital dividend in the 800 MHz band were one of the proposals [http://aikservices.wordpress.com/category/wimax/] [http://www.ofcom.org.uk/consult/condocs/800mhz/800mhz.pdf]
WiMAX makes real-time techniques to operate easily in its bands. Since the carrier is subdivided into multiple subcarriers and if the subcarriers (2048 in total) are used consecutively with respect to frequency or time then this information is known during the allotment phase. One can do horizontal or vertical stripping so as to keep the free segments in one burst as shown schematically in Figure 4. Before the allocation of the channels to mobile stations, the BS can indeed find the available channels/subcarriers and the OFDMA symbol numbers. To acquire the additional spectrum available two WiMAX 17
BS could run in tandem as Master-Slave configuration, which requires some modifications but it is possible. However, more studies are required to find out the feasible implementation of our system with WiMAX. Further, as mentioned earlier it is also possible to work in the TVWS with WiMAX. Figure 4 shows that WiMAX can allocate the spectrum to its users in a particular format so that the remaining carriers could be used by other operators in this case or secondary in general. There is also a possibility of having multiple, WiMAX BS operating in tandem depending on the requirement, i.e., the number of customers and services offered. Within an operatorâ€™s domain, a master BS can redirect a mobile node to a slave BS which comes into being only when there is enough number of customers. The proposed WRAN, IEEE 802.22 also has similar structure.
(b) Figure 4. WiMAX frame diagram (a) with the free OFDMA segments in one burst, and (b) the frame format Source [http://www.wimax.com/commentary/wimax_weekly/2-3-5slot-and-frame-structure] 2) LTE and LTE advanced
C. The Research Issues This section deals with the important challenges that need to be considered to realize a spectrum usage as envisaged in this paper. Let us briefly describe them: 1) There is a need to study in totality the business models -- current and future. 2) Assessing the potential interplay between a Dynamic Licensed regime and a Dynamic License-exempt regime. 3) The specific model of usage of TV white space has to be understood and modeled. 4) Different models of using CR also need to be addressed. For instance OSA can be utilized with least amount of spectrum sensing and with a higher confidence using the CR methods. 5) A thorough study of the current spectrum usage. 6) Policy requirements and solutions are another aspect. How to implement what we propose with respect to the policies of various governments and how to increase the spectrum efficiency. 7) Evaluation of applicable business models is necessary. 8) Development of scalable distributed-systems architecture. 19
9) The basic challenge is to define the functionality and optimization of the allocation system. Having various components, such as policies, value of the services, basic communication needs of the public, influences the allocation system. The cooperation between operators, customers and the allocation system has to be defined. In particular, it should lead to the functional decomposition of the various components of the proposed system (See Section XXX). 10) The design of the basic communication system such as, pilot channels, to support our architecture is another important aspect for building a robust system that is dynamic. 11) Algorithms, protocols and real time and online allocation system are other concerns. Since many frequency bands have to be controlled and allocated the complexity of the system increases. 12) Comprehensive traffic models for various services plays an important role in defining the functionality of various components in our system. 13) Prediction of traffic based on the models and other extraneous information influences the traffic patterns. For example, incidents restricted to geography -like accidents, emergency situation, the time of the year; events such as world-cup games, etc. 14) The important challenge is to define the spectrum efficiency vis-à-vis various services offered and spatial reuse. Since the goal is to increase the spectrum use at a lesser price to the users, defining the basic performance metrics is an important and a difficult challenge. 15) The design of CR equipments that are inexpensive, yet at the same time very flexible capable of changing their radio characteristics over various bands is an important challenge. In fact our methodology assumes this capability and this is a huge challenge for the research community across various areas. 16) Learning systems and self-organization of these systems spatially is another challenge that needs to be addressed if the proposal has to be widely accepted.
X. Conclusions We see spectrum and its usage from a different perspective in this article. This article throws light on the various aspects of the spectrum usage at this point of time. We note that one can do better to increase the usage of the spectrum. We have proposed a model that takes a more pragmatic view of the spectrum usage taking an analogy which most governments indeed look for. We have argued how the spectrum is over-valued or undervalued. Therefore we show that a better model could be devised. Including emergency situations the methodology proposed here could support a more harmonized and natural usage of the spectrum. That implementation is possible with the current technology. To summarize, while there are checks and bounds in our system too, the spectrum is not ‘free’ in the usual sense. We provided a framework where the ‘true value’ of the resource is decided not by one or two actors in a system but everyone contributing to the convergence of the value of the resources to its true value. We have also listed many challenges that have to be addressed before this system. 20