Evolution from Circuit to Packet Switching: The Demise of Circuit Voice

1. The Demise of CircuitVoice SwitchingAndy ValdarPresident of FITCE & Visiting Professor at UCL 4th November 2016

2. So, is TDM digital circuit switching in demise? Leased Lines • Shift from TDM circuits to packet based alternative – Ethernet Voice Switching • ‘Over-the-Top’ services like Skype are IP based • Move to NGN • PSTN replacement by IP-based network • Delay in introducing • 4G/LTE is an all-IP architecture • requires the addition of IMS

3. Circuit Switching Today Digitalcircuit switching is used today to provide voice calls on: • The fixed line telephone networks (Public Switched Telephone Networks) • Mobile networks • 2G (GSM, CDMA, etc) • 3G • 4G (LTE) even!! [although the LTE architecture allows for VOIP solution - VoLTE] • Cable TV networks • Specialised business services networks (Centrex, VPN)

4. Other CS networks GSM and GPRS Voice & data (GPRS) Voice G/MSC MSC Voice BSC Voice Voice & data (GPRS) BC 2G, 2½G Voice PSTN Voice & data 3G Data (GPRS) Voice VoLTE interworking 3G UTRAN GGSN SGSN data data RNC data Circuit- switch fallback voice Voice & data Node B Internet 4G/LTE Data & VoLTE MGW data eUTRAN Data & VoLTE P-GW 4G/LTE S-GW Data & VoLTE Data & VoLTE Other IP networks eNode B data & VoLTE EPC eNode B Key = Packet switched = Circuit switched

5. Why Put Voice Calls Over (IP) Packet Networks? The main arguments made: • It’s cheaper! • Better quality • Circuit-switching is yesterday’s ‘analogue’ technology • Everything is moving to IP anyway Let’s briefly examine these fom an economic perspective........

6. Cost Comparison: Circuit & Packet Switching • Cost rather than price Most people’s perception is influenced by the association of VoIP with free or very cheap calls. • Skype: using broadband access already paid for by subscriber; the Skype company makes almost no profit. • Cheap international calls with specialist service providers. This is price advantage due to by-passing normal voice call interconnect charges between network operators.

7. Cost Comparison: Circuit & Packet Switching 2. No inherent cost difference between the two technologies • Digital. • Both are digital, using a form of PCM encoding of voice • Efficient use of network capacity • Circuit switching uses TDM, channels identified by their timeslot • Packet switching uses interleaving, packets identified by address or label in the header • Technology • Both use high speed VSLI chip technology Digital circuit switches and IP routers look just the same inside the box.... .

8. A TDM T-S-T Switch-Block TS124 TS45 Time Switch C1 A1 Time switch A1 C1 B1 TS10 Time Switch C2 C2 Time switch A2 A2 b2 Space switch B C3 Time switch A3 Time Switch C3 A3 Time switch (in) Space switch Time switch (out) TDM circuit established between A2/TS10 and C1/TS45 via b2:B1/TS124

9. A Packet Router Packet i at t2 Packet i at t3 Buffer #C1 A1 Buffer #A1 C1 B1 Packet i at t1 Buffer #C2 C2 Buffer #A2 A2 b2 Space switch B C3 Buffer #A3 Buffer #C3 A3 Space switch Input Buffers Output Buffers Packet routed from Port A2 (at t1) to Port C2 (at t3) via Space Switch at t2

10. So, there is no inherent cost difference between circuit and packet switches Conclusion 1

11. However, there are 2 architectural features that can give an economic advantage to an (IP) packet switching solution for voice………

12. 3. First Architectural Advantage: Shift of User-Termination Costs

13. Shift of User-Interface Costs • The major costs of voice switching are in the interface cost in terminating the subscribers line(or channel): • 70% of total capital cost of any switching or multiplexing system terminating subscribers’ lines are due to the line cards • True for PSTN subscriber concentrator switches • True for ADSL DSLAMS (IP/ATM) • True for Edge IP routers • Therefore, cost savings can be made by shifting the line terminating functions to the users equipment – customers then pay for it (usually unwittingly) .

14. The Components of a Subscriber Line Card BORSCHT functions: Hybrid 64kbit/s Ringing current Test Overload Battery Supervision Codec Line Card Ring relay & trip detector Line- power feed Over- Voltage protection signalling extractor Ss Test Relay encoder 2Mbit/s decoder s s mux line- feed bus 8kHz test bus Ring bus line unit controller timing

15. Sub’s premises Fixed access network Exchange UNI Shift of User-Interface Costs Operator’s network costs User’s costs MDF Circuit- Switched Network Cu Conc LTE Optical fibre to the home LTU Packet- Switched Network MDF IP phone or Computer with audio mike & speaker Broadband ADSL/Cu DSLAM Conc Broadband/fibre A B C Transmission line termination Subs line card

16. Similarly for Mobile Mobile Circuit-switched network No per-sub equipment Battery, Ringing, & Codec

17. 55th FITCE Congress - Athens 2016 4. Second Architectural Advantage: Multi-Service Platform

18. Multi-Service Platform Concept #1 #n (a) nseparate dedicated platforms #2 (b) single multi-service platform equivalent MSP aggregation & service- interface

19. Multi-Service Platform Concept Multi-Service Platform (MSP) concept offers several advantages compared to the use of separate platforms to support each service., namely: • Total capacity savings, due to combining several services with differing volume distribution throughout the day (i.e. non-co-incident peaks). • Forecasting errors in each of the services create less of a penalty, due to ‘swings and roundabouts’ – i.e. any over-forecasting in some services is compensated by under forecasting in the other services. • Time to market for new services is reduced, since there is already a wide-scale deployment of the platform – so new service is accessible across all of MSP immediately. • Capital cost savings due to single high-volume platform • Operational cost savings due to only one set of technologies to maintain These advantages apply whatever the technology

20. Hence: There are cost savings due to Integration of Voice & Data Traffic onto one platform: • The use of an IP common-services platform (e.g. in Internet or NGN) leads to economies of scale and high utilisation of capacity • High level of capital cost saving is possible due to integration • Early time-to-revenue possible with new product launch (with positive impact on current account income) • Service integrations gives opportunities for new converged features and services – so potential new revenue sources .

21. However, there are difficulties in managing services with a range of QoS – Hard to run differential QoS on same IP platform (without partitioning) • Need to run network at lower than optimum occupancy in order to manage the QoS, especially latency and jitter.

22. Utilisation versus Queueing Delay on IP Platform Source: Capacity Planning for Carrier-Scale IP Networks; Ch 9 of Telecommunications Performance Engineering, The IEE, 2003

23. 5. Other Reasons for VOIP a) New converged services b). New features that exploit the IMS capability • Presence • Location • Multimedia • Machine-machine c). Terminal developments • Screen display capability • Improvements in portable battery life • Integration with other devices (head-up displays, etc) d. There is a general adoption of IP technology, which increases the cost-volume characteristics and so reduces the price of IP equipment (self-fulfilling prophecy!) .

24. The Rationale for Next Generation Networks (NGN) e) Cost Savings Due to Reduction in the number of Existing Network Platforms • The cost savings resulting from the replacement of network platforms by a single NGN platform form the logic for the NGN business case. • Substantial savings in capital costs in equipment to build and expand the NGN platform • Substantial current account savings in running fewer platforms. • Current account savings will only accrue if the old legacy platforms are withdrawn. • In practice it is difficult to completely withdraw network platforms • Difficult business case to construct • Given the practical difficulties in rapid replacement of platforms, particularly when there are reducing revenue streams means the break-even time for the business case can be unacceptably long .

25. Conclusions What should one conclude from this analysis? • On a like-for-like basis there is no inherent cost difference between the circuit-switching and packet-switching technology. • However, packet switching (IP) can benefit from two architectural advantages: • Shift of the user-network interface • Use of a multi-service IP network (e.g: NGN) Hence • When the existing subs line/telephone is kept the NGN business case is impossible because the first architectural advantage cannot be exploited!

26. Conclusions-2 • All mobile systems inherently take advantage of the First architectural advantage (UNI boundary). • The 4G (LTE) architecture is based on a single IP multi-service platform, so once IMS is adopted the ‘Circuit-switch fall back’ will cease and the MSC will be withdrawn.

27. Final Thoughts A Philosophical View: The original concept of an IP network (particularly the Internet) was its simplicity and best-effort performance. However, progressive enhancements over the last few years have made the networks more complicated (protocol stacks) and costly. Much of the enhancement is to impose order and structure on the IP routings by providing virtual paths (e.g. ATM, MPLS, Ethernet) – i.e. making the packet network behave like a circuit-switched network! Perhaps one can foresee a time in the future when….. - the additional costs and complexity to manage the quality of service, plus - the elaborate measures to ensure cyber security …….make circuit switching a more economical choice! !

28. Thank you.