OTT and the future of the PSTN Henning Schulzrinne FCC
PSTN: The good & the ugly The good
The ugly
Global Connectivity (across devices and providers)
Minimalist service
High reliability (engineering, power)
Limited quality (4 kHz)
Ease of use
Hard to control reachability (ring at 2 am)
Emergency usage
Operator trunks!
Universal access (HAC, TTY, VRS)
No universal text & video
Mostly private (protected content & CPNI)
Limited authentication Security more legal than technical (“trust us, we’re a carrier”)
Relatively cheap (c/minute)
Relatively expensive ($/MB) 2
The OTT to “traditional” spectrum
Non-‐interconnected VoIP • Not interconnected
Interconnected VoIP • Bidirectional connectivity to E. 164 numbers • 911 • CALEA • USF
Video relay service • Multimedia for Deaf & HoH • Can reach any E. 164 number via relay
QoS-‐enabled VoIP • [technical possibility] • Can reach any telephone number • QoS as commercial service
Facilities-‐based VoIP • “specialized service” • often, logical, not physical separation (“service flow”) • e.g., MVPD service
user-‐initiated resource reservation (RSVP, NSIS, DOCSIS 3)
Traditional Analog/ TDM POTS • needs no explanation
What are key attributes? * Universality
* reachability à global numbering & interconnection * media à HD audio, video, text * availability à universal service regardless of * geography * income * disability
* affordability à service competition + affordable standalone broadband
* Public safety * * * *
citizen-‐to-‐authority: emergency services (911) authority-‐to-‐citizen: alerting law enforcement survivable (facilities redundancy, power outages)
* Quality * * * *
media (voice + …) quality assured identity assured privacy (CPNI) accountable reliability 4
What is less important? * Technology * wired vs. wireless * but: maintain quality if substitute rather than supplement
* packet vs. circuit * “facilities-‐based” vs. “over-‐the-‐top” * distinction may blur if QoS as a separable service
* Economic organization * “telecommunication carrier”
5
Signaling
Media
Analog
circuit (A)
circuit (A)
Digital
circuit (D)
circuit (D)
AIN
packet (SS7)
circuit (D)
VoIP
packet (SIP)
packet (RTP)
OTT: access to broadband Chart 1 Households With Access to the Fixed B roadband Speed Benchmark by Technology Any Fi xed Cable DSL Fixed Wireless Other Copper Fi ber 0%
20%
40%
6
Eighth Broadband Progress Report, August 2012
60%
80%
100%
Measuring Broadband America
Advertised vs. actual 2012
Chart 1: Average Peak Period and 24-Hour Sustained Download Speeds as a Percentage of Advertised, by Provider—April 2012 Test Data 24 hr Mon Sun
7pm 11pm Mon Fri
140%
Actual/ Advertised speed (%)
120% 100% 80% 60% 40% 20%
ht M ed iac om Qw es Ti m eW t ar Ve ne riz r on Fib Ve er riz on DS W L in ds tre am
sig
r In
Fr on tie
Co x
AT &T Ca bl ev isi on Ce nt ur yL in k Ch ar te Co r m ca st
0%
As shown in Chart 2, upload performance in 7 the April 2012 test data is much less affected than download Bperformance peak periods. Measuring roadband Aduring merica, July 2012 While in 2011 almost all ISPs reached 90 percent or above of their advertised upload rate, in 2012 most ISPs improved to deliver above 100 percent of their advertised rate, even during peak periods.
Significantly better than 2011
Measuring Broadband America, July 2012
Measuring Broadband America
Latency by technology Chart 10: Average Peak Period Latency in Milliseconds, by Technology—April 2012 Test Data
Average Latency (Milliseconds)
Cable
DSL
Fiber
70 60 50 40 30 20 10
M bp s
50
M bp s
35
M bp s
28
M bp s
24
M bp s
20
M bp s
16
M bp s
12
M bp s 8
M bp s 6
M bp s 4
M bp s 2
M bp s 1
0. 51
2
M bp s
0
Advertised Speed (Mbit/s)
9
Measuring Broadband America, July 2012
Charts 11.1-11.5 displays average web page loading 39 time by speed tier for the April 2012 test data period. Web pages load much faster as broadband speed increases, but beyond 10 Mbps,
(a) Download throughput is mostly consistent, with some exceptions.
(a) The biggest difference between peak and worst performance is about 40%.
Other QoS impairments (b) Upload throughput is consistent across ISPs. Figure 7: Consistency of throughput performance: The average throughput of each user is normalized by the 95th percentile value obtained by that user. (SamKnows)
(b) The standard deviation of throughput measurements increases during peak hours, most significantly for ISPs that see lower throughputs at peak hours.
* Packet loss
* VoIP: < 1-‐5% acceptable * Video: loss à lower throughput
“speed test” measurement taken at the wrong time could likely report misleading numbers that do not have much bearing on the long-term performance.
5.3
Effect of Traffic Shaping on Throughput
* Home networks * “Buffer bloat” in gateways
ISPs shape traffic in different ways, which makes it difficult to compare measurements across ISPs, and sometimes even across users within the same ISP. We study the effect of PowerBoost 3 across different ISPs, time, and users. We also explore how Comcast implements PowerBoost.
(c) Loss increases during peak hours for Cox. Other ISPs do not see this effect as much.
* “don’t download that video, Figure 8: Time of day is significant: The average download throughput for Cablevision and Cox users drops significantly durWhich ISPs useI’m PowerBoost, and how does it vary across ISPs? o n t he p hone!” ing the evening peak time. Throughput is also significantly more The SamKnows deployment performs throughput measurements
* Reliability?
once every two hours; each measurement lasts 30 seconds, and each report is divided into six snapshots at roughly 5-second intervals for the duration of the 30-second test (Section 4). This measurement approach allows us to see the progress of each throughput measurement over time; if PowerBoost is applied, then the throughput during the last snapshot will be less than the throughput during the first. For each report, we normalize the throughput in each period by the throughput reported for the first period. Without PowerBoost, we would expect that the normalized ratio would be close
variable during peak time. (SamKnows)
gradual decrease, rather than an abrupt decrease, could be because PowerBoost durations vary across users or that the ISP changes S. Sundaresan et al, Broadband Internet Performance: A View PowerBoost parameters basedACM on Snetwork From the Gateway, IGCOMM state. 2011 From a similar analysis for uploads (not shown), we saw that only Comcast and Cox seem to provide PowerBoost for uploads; we observed a decrease in throughput of about 20%. Dischinger et al. [12] also re-
Broadband virtuous cycle adoption
fixed broadband investment
(relevance, value)
applications (incl. OTT)
OI principles
cellular broadband (backhaul)
broadband availability
Open Internet Principles Transparency. Fixed and mobile broadband providers must disclose the network management practices, performance characteristics, and terms and conditions of their broadband services; No blocking. Fixed broadband providers may not block lawful content, applications, services, or non-‐harmful devices; mobile broadband providers may not block lawful websites, or block applications that compete with their voice or video telephony services
No unreasonable discrimination. Fixed broadband providers may not unreasonably discriminate in transmitting lawful network traffic. 12
Going forward * Interconnected VoIP: done * CALEA, USF, E911 * Part 4 outage reporting
* In progress
* Intercarrier compensation: IP interconnection expectation + transition to bill-‐and-‐keep * NG911, better location * video relay services, CVAA
* To do
* numbering & databases * security model (robocalls, text spam, vishing) * VoIP interconnection model
… , we expect all carriers to negotiate in good faith in response to requests for IP-‐to-‐IP interconnection for the exchange of voice traffic. The duty to negotiate in good faith has been a longstanding element of interconnection requirements under the Communications Act and does not depend upon the network technology underlying the interconnection, whether TDM, IP, or otherwise. Moreover, we expect such good faith negotiations to result in interconnection arrangements between IP networks for the purpose of exchanging voice traffic.