Review of Standardisation Issues for Mobile Network Synchronisation over IP
Ian Wright - Technical Services Manager, Chronos Technology Ltd
Telecoms Networks are ever changing. Migration to Next Generation Networks (NGN) Packet based networks is being planned and in many cases implemented due to their flexibility and cost effectiveness. This migration will give benefits but there is a downside; users of existing service have to be careful that this change does not cause problems with their Quality of Service (QoS). Synchronisation performance is one area where NGN can have a disruptive influence and Mobile Networks are particularly sensitive to Synchronisation performance. At the recent Workshop on Synchronization in Telecommunication Systems WSTS 05 – Boulder, Colorado – 10-12th May 2005, NGN and Mobile Networks were key topics.
At the workshop, experts in the field of synchronization for telecom networks presented papers on and debated this specialist topic. Synchronization impacts telecoms in two major ways. Primarily, it harmonises the way digital communications traffic moves throughout the network ensuring that traffic is neither lost nor overly delayed. Secondly it enables applications at the edge of the network to operate effectively and interwork with the network as a whole.
Opinion is divided amongst the experts concerning the magnitude of the synchronisation problem when Packet Networks dominate, many believe that the migration process will be painful with some existing and critical applications including mobile networks unable to operate or interwork effectively with packet based next generation networks.
Sync for Wireless Networks was examined in detail for GSM, UMTS and CDMA synchronization. Charles Curry of Chronos Technology Ltd, UK reviewed the current standards for GSM and 3G/UMTS networks which referenced sync requirements and proposed an MTIE quality mask for a Node B base station. He warned that equipment manufacturers do not understand the implications of poor sync and poor architecture and mistakes have been made in the past (and continue to be made today) including synchronizing base stations to E1/T1 traffic feeds rather than dedicated sync feeds. He also highlighted the need to test network equipment to see if it is fit-for-purpose for sync delivery, particularly the new range of IP access products aimed at NGN networks. Finally Charles observed that there are no standards for sync quality or sync availability at the base station and that with the new generation of sync over packet solutions, what confidence can operators have in wide scale deployment without significant testing under challenging conditions?
Barry Dropping of Symmetricom showed evidence of how deploying improved synchronisation in GSM Networks had significantly reduced the dropped call rate and in one trial deploying synchronisation at Base Station Sites had increased the time between dropped calls by 25%. He further illustrated how poor sync can be linked to poor quality of service in a mobile network and better sync will reduce churn.
Synchronisation is not always considered sufficiently during the network design phase, operators continue to sync their base stations from the E1 or T1 line clock when this was prone to phase hits from SDH/SONET pointers despite sync experts' warnings.
Jeremy Lewis of Zarlink, Canada. focussed on Timing & Sync over packet networks, outlining the different mechanisms for transporting clock across a packet network, including “In” and “Out” of band clock recovery and “Adaptive”, “Differential” and “Combination” clock recovery. Key quality affecting features are packet delay variation, network loading and network complexity. As these parameters increase, they adversely impact sync quality. The inclusion of better oscillators helps to improve things in the case of “Adaptive” and “Combination”, and network clock needs to be present through some other transport mechanism with “Differential”.
Pat Diamond of Semtech, USA introduced IEEE-1588 an emerging Two Way Timing Transfer protocol for sync delivery. Results were shown of good performance over a Trial configuration on a live Network. Since no standardised testing procedures have yet been defined to examine the performance of sync transport over congested as well as uncongested networks, some delegates were sceptical as to the applicability of this technology for carrier class sync transport in an Ethernet environment. It was agreed that views might change given further testing.
An excellent paper from Mike Gilson, BT, UK expanded on the unpredictability of testing processes on manufacturers' equipment. Now that BT is committed to NGNs with their 21CN programme, it is vital to qualify access layer equipment as fit-for-purpose to transport sync. Unfortunately results are not consistent. The variation between different manufacturers' equipment for a range of payload sizes was significant, some working very well, others performing well outside the G.823 wander standard. The repeatability envelope of MTIE results showed that the same test, under the same loading conditions can produce considerably differing results. Worrying conclusions with a call for harmonised testing standards and summarised by the statement that End to End sync transparency has “Some way to go”.
Michel Ouellette's, Nortel, Canada, paper, one of the best papers of the conference, was a Quantitive Study of Timing over Packet Networks. Nortel had evaluated a number of different manufacturers' Circuit Emulation solutions in an Ethernet environment to interconnect PBX's. Again in a similar conclusion to Mike Gilson, some results were quite disturbing, showing product that would just not work! There were some very long learning cycles at the beginning of the test and in one case after a “Link Failure” simulation; one solution had to be power cycled in order to continue to operate. Michel clearly underlined that fact that understanding packet based network impairments is not sufficient, without considering the needs of the legacy (existing) products and services.
The final paper of the conference was delivered by Silvana Rodrigues of Zarlink, Canada and reviewed Standards work on sync over packet. The four key areas of activity are IEE-1588, NTP, ITU SG15/Q13 and OPTXS.SYNC. A number of delegates present at the conference meet again in Geneva the following week as part of the ITU SG15/Q13 group working on the ITU recommendation G.Pactiming.
The conference wound up with an extended panel session debating the main issue of sync over packet or “Next Generation Networks”. The panel agreed that the industry was entering a period of uncertainty with claims of suitability by companies with sync over packet solutions juxtaposed with some quite poor test results. The industry needs Sync Standards for circuit emulation over IP which are not defined as yet and these will need to be complemented with testing procedures and availability rules.
In short, the conference confirmed that no Standards exist yet for transporting sync over a packet network, no methodology has been defined for measuring the quality of sync transport over a packet network or of defining service level agreements between mobile operators and backhaul providers, and much uncertainty exists as to the quality of sync necessary for effective operation of a mobile network. Clearly an interesting state of affairs given the amount of investment in mobile networks.
The conference closed with a call to attend the International Telecom Sync Forum at the IEE in London October 17-19th 2005. Quite a few of the presenters at WSTS will also be presenting at ITSF 05.
Acronyms
21CN - Twenty First Century Network E1 - 2.048Mbit/s Interface ITSF - International Telecom Sync Forum MTIE - Maximum Time Interval Error NGN - Next Generation Networks NTP - Network Timing Protocol OPTXS - Optical Transport and Synchronisation Committee (formerly T1X1) PBX - Private Branch Exchange QoS - Quality of Service SDH - Synchronosus Digital Hieracht SONET - Synchronous Optical Network T1 - 1.544Mbit/s Interface WSTS - Workshop on Synchronization in Telecommunication Systems
|