+ All Categories
Home > Documents > UMTS Synchronisation

UMTS Synchronisation

Date post: 24-Oct-2015
Category:
Upload: dclaudel
View: 37 times
Download: 7 times
Share this document with a friend
Description:
Book
30
umts synchronisation.doc v1.31 Page 1 of 30 08 Feb. 06 UMTS Network Synchronisation Charles W T Curry B.Eng. (Hons) – Chairman - Chronos Technology Ltd www.chronos.co.uk Introduction Mobile Operators are one of the most powerful customer groups with whom Network Operators have to develop commercial relationships. Business decisions taken in the early stages of UMTS/3G rollout will undoubtedly have a significant impact on the medium term viability of the business. The right partnerships need to be developed. Wireless operators are in the process of making a massive capital investment in UMTS licences and infrastructure. If the technology is to deliver increased speed over GSM, will the lessons learned from GSM be heeded, particularly regarding synchronisation? Now is an ideal opportunity to get it right. Additionally, services to be offered over UMTS include location services. How will these services be affected by poor synchronisation at the base stations? This paper explores the synchronisation requirements for UMTS networks and offers solutions for Core Network and Node B synchronisation, with suggestions for quality metrics. It also illustrates types of synchronisation problems experienced by the author when measuring real networks and explores the consequential effect on Node B behaviour. Mobile Operators’ Perspective From the “synchronisation-aware” Mobile operators’ perspective, the questions most likely to be to be asked should be: - Q. Can the Network Operator demonstrate why synchronisation is required for the successful operation of my UMTS/3G network and backhaul of my Node B traffic? Q. How can the service being offered to our end customers, be enhanced and protected over and above simple local traffic delivery? The quantity of backhaul bandwidth for UMTS/3G networks is likely to be relatively large given that 3G networks are predicted to have between 3 and 6 times more Node B’s than Base Stations in the 2G GSM world. If this is true then the business opportunity to suppliers of E1 access connectivity and significant core bandwidth
Transcript
Page 1: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 1 of 30 08 Feb. 06

UMTS Network Synchronisation

Charles W T Curry B.Eng. (Hons) – Chairman - Chronos Technology Ltd www.chronos.co.uk

Introduction Mobile Operators are one of the most powerful customer groups with whom Network Operators have to develop commercial relationships. Business decisions taken in the early stages of UMTS/3G rollout will undoubtedly have a significant impact on the medium term viability of the business. The right partnerships need to be developed. Wireless operators are in the process of making a massive capital investment in UMTS licences and infrastructure. If the technology is to deliver increased speed over GSM, will the lessons learned from GSM be heeded, particularly regarding synchronisation? Now is an ideal opportunity to get it right. Additionally, services to be offered over UMTS include location services. How will these services be affected by poor synchronisation at the base stations? This paper explores the synchronisation requirements for UMTS networks and offers solutions for Core Network and Node B synchronisation, with suggestions for quality metrics. It also illustrates types of synchronisation problems experienced by the author when measuring real networks and explores the consequential effect on Node B behaviour. Mobile Operators’ Perspective From the “synchronisation-aware” Mobile operators’ perspective, the questions most likely to be to be asked should be: - Q. Can the Network Operator demonstrate why synchronisation is required for the successful operation of my UMTS/3G network and backhaul of my Node B traffic? Q. How can the service being offered to our end customers, be enhanced and protected over and above simple local traffic delivery? The quantity of backhaul bandwidth for UMTS/3G networks is likely to be relatively large given that 3G networks are predicted to have between 3 and 6 times more Node B’s than Base Stations in the 2G GSM world. If this is true then the business opportunity to suppliers of E1 access connectivity and significant core bandwidth

Page 2: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 2 of 30 08 Feb. 06

transmission capacity in providing the mobile operator with backhaul bandwidth may well be many hundreds of Millions of Euros. The 3G mobile operator must ensure that their chosen bandwidth supplier can meet the very exacting standards that are demanded for this solution. This local E1 Node B connectivity and backhaul bandwidth does after all contribute significantly to their business case and future success! A Network Operator who can demonstrate that they understand the need for synchronisation, can offer ways of optimising the local access infrastructure and can offer opportunities to enhance end customer offer, is the ideal partner for a 3G mobile operator. Network Operators’ Perspective A Network Operator that wishes to be the supplier of E1 Node B access connectivity and core transmission capacity to a 3G Mobile Operator, has to ask a few simple questions. Q. Am I able to guarantee the quality of synchronisation delivery through my local access network such that the jitter and wander characteristics meet the requirements of the UTRAN? Q. Can the local access architecture be simplified and made more cost effective? Q. What additional services may a prospective Mobile Operator require at remote Node B locations and can I offer a solution to meet those needs? Mobile Operators are looking to award significant business (often hundreds of Millions of Euro’s) with Network Operators that understand their remote site needs. A Network Operator who can demonstrate that they understand the local access architecture options and the need for synchronisation and can offer the “value-add” of guaranteed synchronisation will be in a strong position to win UMTS/3G backhaul network solutions. The UMTS Network and Synchronisation In order for Universal Mobile Telecommunications Services (UMTS or 3G) to work correctly, the absolute radio frequency (RF) stability of air-interface transceivers - Node B’s in the UMTS Terrestrial Radio Access Network (UTRAN) must be maintained to within defined standards.

Page 3: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 3 of 30 08 Feb. 06

Traditionally, the air interface has derived its RF stability from the incoming link of the telecom network used to deliver traffic with the base stations. Whilst the long term stability of these E1 links is often beyond question since they are directly traceable to a network Primary Reference Clock (PRC), the short and medium term variation in frequency (wander) may have a significantly detrimental effect on the air interface stability. Whilst some manufacturers are now designing low cost OEM Global Positioning (GPS) receivers into their Node B’s, risk of loss of GPS reception means that wireless network operators will still need to observe and make use of the E1 frequency stability information. If the relevant standards are not met, the ability of the network to operate effectively will be adversely affected. Poor synchronisation in GSM networks is known to compromise cell handover particularly whilst calling from a moving platform; it can also be a cause of dropped calls whilst stationary. Mapping of E1 traffic though the Synchronous Digital Hierarchy (SDH) can cause pointer movements (VC12 pointers). This is known to cause base stations to temporarily cease functioning. The purpose of this paper is to analyse the synchronisation issues of the UTRAN. It will also propose synchronisation quality requirements in terms of metrics familiar to telecom network operators and define sync transport mechanisms which guarantee sync quality delivery from the core network to the access layer and Node B’s. Relevant Standards and Metrics The main standards for telecom network synchronisation are the European ETSI Standards series EN 300 462-1 to EN 300 462-7, International Telecom Union ITU Series G.811, G.812 and G.813 and the American T1X1 T.101. There is a lot of similarity between the language in the three standards’ bodies and in all, the metrics for telecom network synchronisation stability are now well established and very well defined in terms of Maximum Time Interval Error (MTIE) and Time Deviation (TDEV). References to synchronisation in the standards defining requirements for 3G networks can be found in many documents within the 3rd Generation Partnership Project (3GPP) family. However the technical specification capturing most of the requirements is TS 125 402 – “Synchronisation in UTRAN Stage 2”. References to other relevant 3G technical specifications can be found within TS 125 402. The classical specification for the RF accuracy of GSM base stations was .05 ppm or 5x10-8 and is found in ETSI TS 145 010 (TS 100 912). The equivalent UMTS specification can be found in TS 125 104 “UMTS: UTRA (BS) FDD; Radio

Page 4: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 4 of 30 08 Feb. 06

transmission and reception” and TS 125 105 “UMTS: UTRA (BS) TDD; Radio transmission and reception” but modifies the accuracy to ±5x10-8 over one timeslot. We can also find in TS 125 402 a specification for the relative phase difference between Node B’s of 2.5 µs. Let’s briefly review what the 3GPP specifications say about synchronisation: - ETSI TS 125 402: “UMTS: Synchronisation in UTRAN Stage 2” states that: - “A general recommendation is to supply a traceable synchronisation reference according to G.811. The clock to be implemented in UTRAN Nodes shall be chosen with characteristics that depend on the L1 adopted (see TS 125 421 and TS 125 431) and on the Network Synchronisation strategy adopted. Already standardised clocks may be used (see G.812, G.813, EN 300 462-4-1, EN 300 462-5-1 and EN 300 462-7-1). For example in order to support STM-N interfaces at the RNC, the ITU-T G.813 may be sufficient. The implementation in the UTRAN of a better performing clock (in terms of holdover) may be recommended for distribution of a 0.05 ppm during failures in the synchronisation network (EN 300 462-7-1, EN 300 462-4-1, or ITU-T G.812 type 1, type 2 or type 3).” ETSI TS 125 411 “UTRAN Iu Interface layer 1” sub clause 4.2 states “The jitter and wander performance requirements on the interface shall be in accordance with either G.823, G.824, G.825, whichever is applicable. The synchronisation reference extracted from the Iu may be used as UTRAN synchronisation reference. A general recommendation is to supply a traceable synchronisation reference according to G.811.” ETSI TS 125 104 “UTRA (BS) FDD & TS 125 105 “UTRA (BS) TDD: Radio transmission and reception”, sub clause 6.3 “The modulated carrier frequency of the BS shall be accurate to within ± 0.05 ppm observed over a period of one power control group (timeslot).” Wherever the synchronisation reference originates – network PRC or local GPS, this synchronisation information must be passed through the UTRAN in order to ensure that all clocks within the UTRAN, whether they be at RNC sites or Node B sites meet the requirements for correct operation. As already stated, Node B’s must have frequency accuracy of ±5 x 10-8 over one timeslot. This means that any frequency reference used by the UTRAN and ultimately delivered to the Node B must meet this accuracy at all times. From this perspective

Page 5: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 5 of 30 08 Feb. 06

TS 125 402 recommends G.811 (EN 300 462-6) traceable references for the whole UTRAN. Whilst the Core Switch sites and co-located Radio Network Controller (RNC) are most likely to interface to a transmission network at the STM-N rate and only theoretically needs to meet the SEC standard of G.813 or ETSI EN 300 462-5, two key issues must be considered. These are (i) are we exceeding the overall SDH network architecture with reference to SEC clock hop count, which is defined as not exceeding 20 in ETSI EN 300 462-2-1? and (ii) the need to maintain Node B frequency accuracy in the event of holdover at the RNC. This therefore implies that, realistically, every Core Switch site should be fitted with a fault tolerant SSU incorporating references meeting G.812 or EN 300 462-4-1. These can be further enhanced with local GPS timing receivers to flatten the synchronisation architecture with a local primary reference source. SSU’s should be of the multiple input variety with local quality monitoring. In this way any interconnects with third party backhaul carriers can be monitored as well. Requirements for Location Services It is becoming a requirement in the USA to locate mobile 911 (emergency) calls. It is planned to offer location services on UMTS networks. Two standards, which started life in the GSM domain and have been updated for UMTS are relevant. These are TS 03.71: TS 101 724: “Digital cellular telecommunications system (Phase 2+); Location Services (LCS); Functional description; Stage 2” and TS 45.010: TS 145 010: “Digital cellular telecommunications system (Phase 2+); Radio subsystem synchronization”. LCS utilises one or more positioning mechanisms in order to determine the location of a Mobile Station (MS). Three positioning mechanisms are proposed for LCS: Uplink Time of Arrival (TOA), Enhanced Observed Time Difference (E-OTD), and Global Positioning System (GPS) assisted. The TOA method measures the time of arrival of a known signal sent from the MS and received at three or more measurement locations. The relative time difference between any two measurement locations is computed centrally and the position of the MS calculated by hyperbolic trilateration. In order to enhance the TOA method, Location Measurement Units (LMUs) can be used whose coordinates are accurately known. These LMUs are fixed receivers which can communicate error information to the central controller. The E-OTD method is based on measurements in the MS of the Enhanced Observed Time Difference of arrival of bursts from nearby pairs of Node Bs. To obtain accurate triangulation, E-OTD measurements are needed for at least three distinct pairs of geographically dispersed Node Bs. Based on the measured E-OTD values the location

Page 6: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 6 of 30 08 Feb. 06

of the MS can be calculated either in the network or in the MS itself, if all the needed information is available in the MS. LMUs can also be used in this method. Location services utilising the triangulation concept, as anyone with an interest or background in navigation will know, require that the relative time difference between the transmitter stations be as small as possible since this becomes critical to obtaining good positional accuracy. If the Node Bs are getting their air interface stability from the incoming E1, and if this is subject to wander as described above, then it is possible that their relative positions with respect to each other are changing. We need to examine what will be required from a pragmatic perspective and what can reasonably be achieved. The speed of light is approximately 300,000 km/sec. A time error of 100ns equates to a distance error of 30m – arguably the maximum acceptable error for a useable location service. 100ns error shared between two based stations becomes 50ns per base station. For location services to work effectively all the Node B’s in the network will need to have a relative phase stability of 50ns to UTC. It is not clear yet how different infrastructure or handset manufacturers will address the issue of error correction. However it is clear from the requirements of Annex C of TS 145 010 which details relative phase stability requirements for adjacent base stations that the air interface stability will need to be maintained and monitored with a high degree of accuracy over significant periods of time for location services to offer positional accuracies better than 30m. UMTS Radio Access Network (UTRAN) Synchronisation UMTS network synchronisation can be subdivided into five categories:- 1) Overall Network Synchronisation – the distribution of synchronisation within the

UTRAN and its integration within connecting networks. 2) Node Synchronisation – the relative timing differences between Radio Network

Controllers (RNC’s) and Node B’s and between Node B’s. 3) Transport Channel Synchronisation – Frame transport between the RNC and the

Node B’s. 4) Radio Interface Synchronisation between the Node B’s and the User Equipment

(e.g. mobile phone or modem). 5) Time Alignment handling between the Core Network and the UMTS Radio

Access Network (UTRAN).

Page 7: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 7 of 30 08 Feb. 06

RNC RNC

Node B

Node B

Node B

Node B

Node B Node B

Node B

UserEquipment

UserEquipment

RadioNetwork

Subsystem

UTRAN

CoreNetwork

RadioInterfaceSynchronisation

TransportChannelSynchronisation

TimeAlignmentHandling Iur

Iu

Iub

Figure 1: UTRAN Synchronisation Architecture

The basic synchronisation architecture shown in Figure 1 is defined in ETSI TS 125 401 “UTRAN Overall description” subclause 9 – Synchronisation. For the purpose of this paper we will focus on (1) and (2) above. Identifying suitable synchronisation models will then ensure that (3), (4) and (5) will work correctly. The following sections examine each of these synchronisation categories in turn and offer ideal synchronisation solutions and model architecture for 3G wireless network planners. Overall UTRAN Network Synchronisation In general terms because UMTS elements are interfacing into standard telecom SDH networks designed to meet current requirements, all ETSI and ITU standards should be met at any E1 or STM-N interfaces. In fact TS 125 402 recommends all references meet the relevant standards particularly ITU-T G.812 (SSU), G.813 (SEC), ETSI EN 300 462-4-1 (SSU), EN 300 462-5-1 (SEC) and EN 300 462-7-1 (Local SSU) being the appropriate SEC or SSU standards. Figure 2 below shows the UTRAN network from a more practical perspective. Individual metro Node B’s will backhaul their traffic via E1’s into the access layer, passing over appropriate E1 network terminating equipment into an SDH multiplexer.

Page 8: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 8 of 30 08 Feb. 06

Groups of rural Node B’s will combine traffic into an STM-1 SDH link spur. All this traffic will be groomed into the core network and then delivered to the RNC over an STM-N link of appropriate bandwidth.

Node B

Node B

Node B

Node B

MUX RNC

3rd PartyCore Network

AccessNetwork

Node BNode B

Node B

Node B

AccessMUX

MUX

MUX

MUX

3rd PartyBackhaul

Carrier

E1

STM-1

STM-N

Node B Node B

Node B

CoreSwitch

Site

Figure 2: UTRAN Network

Based on the observations and recommendations of TS 125 402, Figure 3 provides a model method for synchronising Core Switch and co-located RNC sites. The SSU should be capable of observing and using multiple reference inputs. One will be a local GPS; one could optionally be a local caesium to provide the wireless operator with his own G.811/EN 300 462-6 traceability in the event of GPS failure. Two further inputs would be from the line clock of the wireless operator’s own SDH core backbone network; one from the SDH network multiplexer line clock out of the “line east” direction and one from “line west”. These would be for back-up purposes in the event of loss of local GPS, network architecture and planning recommendations of ETSI EG 201 793 must be observed with protection from potential sync loops.

Page 9: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 9 of 30 08 Feb. 06

Backhaul Carrier A

SSU

GPS Receiver

Backhaul Carrier B

Line Clock fromSTM-N MUX

Own NetworkLine West

Reference FrequencyOutputs meeting EN 300462-4 and traceable toG.811/EN 300 462-6

Optional CaesiumPRC

Line Clock fromSTM-N MUX

Own NetworkLine East

Figure 3: Core Switch Synchronisation Method

Some wireless operators will rely on third party backhaul carriers to bring traffic into the Core Switch site from the Node B’s. Utilising an SSU with additional reference inputs will allow monitoring of the synchronisation from these carriers’ networks. Whilst these inputs should be marked for monitoring only and not used as valid synchronisation references, they do offer additional independent references which will serve to confirm the quality of the wireless carrier’s own network. The ideal SSU to meet this model will be fully redundant architecture, centrally managed with at least six reference inputs (more if more backhaul carriers are used) and with MTIE monitoring on all inputs. This model will provide a solid foundation upon which to build the synchronisation platform for the UMTS network. It will meet all the recommendations put forward in TS 125 402 subclause 4.2 Network Synchronisation.

Page 10: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 10 of 30 08 Feb. 06

Access/RNC Synchronisation Access Nodes and RNC’s control multiple Node B’s via line of sight or the radio network. Often the RNS’s are co-located with the Core Switches, but sometimes they are remotely deployed, particularly for geographically diverse networks. By definition access Nodes are remotely deployed, they will usually be rural high sites such as tall masts or hilltop locations with a panoramic view of nearby Node B’s. Transmission links for Access Nodes is usually by local access SDH since the bandwidth required for up to 30 Node B’s makes this technology the most appropriate. The recommended solution for Access Nodes and remote RNC’s is a single GPS system offering sufficient holdover stability to keep all linked Node B’s at better than 5 x 10-8 stability in the event of loss of sync from the network or GPS. Redundancy is not essential since a bypass route from the network to the Access /RNC Node should be designed in. Remote management and the ability by the wireless operator to monitor any 3rd party backhaul synchronisation is strongly recommended. The ideal Access/RNC synchronisation solution (Figure 4) will be a relatively low cost telecom GPS unit offering up to (say) 10 frequency outputs for connecting to co-located network equipment. It will also have the ability to monitor incoming 3rd party sync links or the outgoing sync from the wireless carrier to ensure that sync quality is known at all times.

Backhaul Carrier A GPS Backhaul Carrier B

Reference Frequency Outputsmeeting EN 300 462-4 and

traceable to G.811/EN 300 462-6

Figure 4. Access/RNC Synchronisation Method

Page 11: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 11 of 30 08 Feb. 06

Node Synchronisation Node synchronisation refers to the achievement of a common timing reference amongst different nodes and is defined to ensure that Node B’s are traceable to the RNC and to minimise cross interference between Node B’s. The Node synchronisation method depends on whether a Frequency Division Duplex (FDD) or Time Division Duplex (TDD) mode is being deployed within the UTRAN. FDD mode uses different transmit and receive frequencies for the air interface, TDD mode uses the same frequency but different timeslots for both transmit and receive. FDD and TDD modes have different timing requirements regarding the accuracy of the timing difference estimation and on the necessity to compensate for these differences. In FDD mode a single timing reference is not actually necessary within the UTRAN and phase alignment between Node B’s is not essential. However Node B’s should all be traceable to the same long-term frequency reference (e.g. Universal co-ordinated time (UTC) with the GPS of satellites) However in order to minimise transmission delays and buffering time it is useful to estimate phase or timing differences between the RNC and Node B’s. In TDD mode, the achievement of a common timing reference among Node B’s must be used to support cell synchronisation. The two types of node synchronisation are respectively: - • “RNC-Node B Node Synchronisation” for FDD mode UTRAN • “Inter Node B Node Synchronisation” for TDD mode UTRAN.

RNC-Node B Node Synchronisation If an accurate reference timing signal (e.g. local GPS or PRC traceable source) is available at both the RNC and Node B, the frequency deviation between nodes will be low. However, if a frequency offset exists at the access layer boundary with the Node B, this will cause the RNC-Node B synchronisation procedure to frequency-lock the Node B to the RNC in order to minimise the frequency deviation between Node B’s in the RNC area i.e. Radio Network Subsystem (RNS). The RNC-Node B Node Synchronisation procedure finds the round trip delay for the dedicated channel being used by the user equipment and uses this information to compensate for any frequency offset.

Page 12: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 12 of 30 08 Feb. 06

This system may not work so well where Node B’s are deriving their timing from an incoming E1 which is experiencing significant wander due to playout buffer activity. This wander is usually cyclical and may well compromise the working of the RNC-Node B Node Synchronisation algorithm. It would seem therefore that the quality of the incoming E1 in terms of MTIE will be critical to good operation of the RNC-Node B Synchronisation procedure. Inter Node B Node Synchronisation Inter Node B Node Synchronisation is used to achieve a common timing reference between Node B’s in order to minimise cross interference. The TDD mode is probably more critical of poor synchronisation than the FDD mode because the transmit and receive channels share the same RF frequency and can exist in adjacent timeslots. Inter Node B Node Synchronisation can be achieved in two ways – (i) by using an external sync reference input on the Node B to connect to a local GPS or other suitable reference derived from the incoming transmission link and (ii) over the air interface. Node B’s can also be daisy chained so that one reference could supply a number of Node B’s in for example an indoor application where electrical connectivity between Node B’s is possible. The Inter Node B Node Synchronisation process has a commissioning (non traffic) phase leading to a steady state (traffic) phase. It requires at least one cell in the RNS to be synchronised by an external reference (e.g. GPS receiver). Initially (in the “Preliminary Phase”) the RNC uses the timing from the externally clocked cell to adjust all the timings of the cells in the RNS. Then in the “Frequency Acquisition Phase” the reference cell transmits cell sync information which is picked up by the other cells in the RNS, these cells need to be locked to within 50ppm of the reference cell before this phase is complete. The final part of the commissioning process is the “Initial Synchronisation Phase”. In this phase, individual cells listen for transmissions from other cells and adjust their relative synchronisation accuracy until network requirements are met. Eventually the “Steady State Phase” is reached which continues throughout the operational lifetime of the Node B. This process ensures that the requirements of TS 125 402 subclause 6.1.2.1 are met. TS 125 402 subclause 6.1.2.1 states “The relative phase difference of the synchronisation signals at the input port of any Node B in the synchronised area shall not exceed 2.5 µs.”

Derivation of Node B MTIE If the Node B is to be synchronised from the incoming transmission link rather than a local GPS, a suitable MTIE mask should be created so that the relevant equipment

Page 13: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 13 of 30 08 Feb. 06

and transmission paths used can meet the requirements for UTRAN and Node B synchronisation. Let’s start with the G.811 PRC traceability statement from TS 125 411 subclause 4.2 “Layer 1 Description” and use the ETSI mask for Network PRC limits from EN 300 462-3-1 subclause 7.2.1 Figure 3: “Network limit for wander at PRC outputs expressed in MTIE”. This provides our foundation. Next let’s examine the frequency accuracy statements from TS 125 104 and 105 subclause 6.3. A maximum frequency error of ±5x10-8 at the air interface must equate to an absolute maximum allowable offset of 5x10-8 for each Node B. Putting some margin for error of (say) approximately 20% reduces this to 4x10-8 which equates to a phase error of 40 ns over 1 second and 400 ns over 10 seconds. The MTIE of a constant frequency offset is a straight line, we can extrapolate this line downwards until it cuts the PRC mask from EN 300 462-3-1. In order to obtain the upper limit of the 5x10-8 MTIE let us now examine TS 125 402 statement from subclause 6.1.2.1. regarding the requirement for a phase difference of 2.5 µs between Node B’s, individually each Node B must be half of this from absolute i.e. 1.25 µs. It would seem reasonable to put some margin for error, so rounding to 1.00 µs is proposed. This means that the maximum phase error or MTIE must not exceed 1.00 µs at least until the EN 300 462-3-1 limit us reached over longer observation periods. The 1.00 µs ceiling will also provide an upper limit for the 5x10-

8 slope. We can now create the MTIE mask for a Node B. Since the 2.5 µs specification is only mentioned in the TDD aspect of the UTRAN and then only for Node B’s daisy chained, there is some thought that this may not refer to a relative phase offset for Node B’s in general. For comparison purposes, therefore, the 5 x 10-8 MTIE is included. Additionally, the G.823 Table 2 specification for the allowable wander at 2.048 kbit/s outputs is shown. This clearly indicates that terrestrial network E1 feeds conforming to G.823 would not be suitable for synchronisation feeds for Node B’s unless the Node B has significant wander filtering or the E1 was retimed to a suitably wander free local synchronisation source.

Page 14: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 14 of 30 08 Feb. 06

Figure 5: Node B MTIE compared to Network SEC and PDH MTIE Masks

E1 (Iub) Delivery in the Last Mile Let us now examine three different E1 (Iub) delivery media and the resultant wander effects. These are: 1) ATM E1 circuit emulation – playout buffer wander. 2) E1 delivered over SDH with PDH at the last mile access – buffer and pointer wander. 3) SDH – VC12 pointer wander.

Page 15: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 15 of 30 08 Feb. 06

Figure 6: Phase of E1 from ATM Circuit Emulation Figure 6 shows about 30 minutes of phase data from the playout buffer of an E1 circuit emulation function from an ATM switch. This switch was synchronised to a local GPS receiver. Despite local clocking, because the clock of E1 being passed through was varying, the playout clock was changing resulting in the phase changes observed and introducing wander onto the E1. So even though the E1 is traceable to GPS in the long term, its short to medium term stability is seriously compromised. This particular test was conducted on an E1 potentially being deployed as the link for a GSM base station and was shown to be not fit for purpose.

Page 16: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 16 of 30 08 Feb. 06

Figure 7: ATM E1 Circuit Emulation showing Frequency Variation Figure 7 shows the same ATM E1 circuit emulation service and its effect in the frequency domain. Note that the frequency regularly goes outside the 5x10-8 specification as shown in the expanded view Figure 8.

Figure 8: Expanded view of Figure 7

Page 17: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 17 of 30 08 Feb. 06

Figure 9: Phase of E1 ISDN PDH link delivered over SDH and PDH. Figure 9 shows just over 21 hours of phase data from an E1 ISDN link delivered over a mixture of SDH and PDH via fibre to the . Significant buffer wander is shown which is caused by normal desynchroniser activity and traceable back to the local switch. The other event worth noting is the VC12 pointer. Figure 10 shows the effect that a VC12 pointer has on the phase of an E1 delivered out of a SDH Multiplexer.

Figure 10: E1 out of SDH Network with VC12 pointer

VC12 Pointer

Buffer Induced Wander

Page 18: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 18 of 30 08 Feb. 06

In isolation each of the phase plots shown in Figures 6, 9 and 10 are difficult to quantify, however, when the corresponding MTIE graphs are computed and overlaid against the previously proposed Node B MTIE mask shown in Figure 11 below, a comparison can be made and the conclusion drawn that none of these E1 delivery media would be fit for purpose for passing synchronisation information to a Node B if the Node B has no wander attenuation capability.

Figure 11: MTIE_1 - E1 from ATM, MTIE_2&3 E1 from SDH/PDH, MTIE_4 - E1 with VC12 Pointer

Close examination of the MTIE graphs and masks above reveals the following:-

• Table 2/G.823 – Network Limits for the output wander at a 2.048 kbit/s traffic interface is well in excess of the 5 x 10-8 MTIE up to an observation interval of about 350 seconds.

• VC12 pointers (Plot 4) are well outside the 5 x 10-8 MTIE. • An E1 from an ATM circuit emulation service is outside the 5 x 10-8 MTIE.

Now, it is possible that the Node B manufacturers have built into the input stage of the Node B some kind of relative phase attenuation which might attenuate cyclical phase variations as shown in Figures 6 and 9 and shown in Plots (1) and (3) above. However since in the UTRAN there is no need to phase align RNC and Node B counters, this may not be the case. In addition VC12 pointers, which occur when the VC12 is passed between SDH networks that have non-identical clocks, are a more catastrophic form of wander and less likely to be able to be attenuated without significant filtering involving more expensive oscillator components in the clock

Page 19: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 19 of 30 08 Feb. 06

recovery section of the Node B – something which cost sensitive and competitive Node B manufacturing strategy may not include within the design brief. In some GSM base stations VC12 pointers have been shown to cause the base station clock recovery circuitry to go out of lock for long periods as shown in Figure 12 below.

Figure 12: VC12 Pointer test on GSM Base Station - Frequency Plot At the onset of the VC12 pointer the frequency of the base station suddenly moved to approximately 3.5x10-7 which is well outside the 5x10-8 requirement and did not fully recover until after nearly 4 minutes. Hopefully base station manufacturers have learned the lessons of the damage that a VC12 pointer can cause. Key Requirements for 3G – UMTS Wireless Operators So the 3G wireless operator must ensure that each of the following is in order whilst assessing his Node B connectivity architecture: a) E1’s used for traffic delivery to the Node B should be of a suitable quality to

allow the desired relative phase error to be met – this may be relatively easy to achieve when the wireless carrier owns the E1’s but less easy to guarantee when leasing bandwidth from a third party carrier.

b) Node B’s must be capable of withstanding VC12 pointers. c) Node B’s must be capable of being interfaced to poor quality E1 feeds. Having created model architecture for Core Switch and Access/RNC synchronisation, it is now appropriate to create model architecture for access layer synchronisation delivery to Node B’s. If local GPS receivers are not planned to be co-located with each Node B, a mechanism must be found which guarantees delivery of PRC

Page 20: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 20 of 30 08 Feb. 06

(G.811/EN 300 462-6) traceable synchronisation across the access layer from the core network. The SDH multiplexer at the access layer network site delivering E1’s to the remote Node B’s will have a line clock output traceable to the network PRC and conforming to G.813/EN 300 462-5. The MTIE mask of this synchronisation reference is within the 1.25 µs requirements identified above and therefore suitable if it is not degraded by the E1 communication link. Optionally a local SSU or GPS could be co-located with the multiplexer, this would not only enhance the network site to G.812/EN 300 462-4/7 but also allow the MTIE of the SDH line clock to be measured and monitored back at the network management centre. The better holdover stability would ensure that Node B synchronisation was maintained even if the SDH multiplexer lost its traceability to the network PRC. The E1 delivery to the Node B is still at risk from VC12 pointers. The Terminating Element must be able to retime the E1 to the SDH line clock and deliver the sync to the far end with sufficient quality to meet the Node B requirements.

Figure 13: Comparison of SSU, SEC and PRC MTIE Masks with Node B requirement and GPS

Receiver

Figure 13 shows that if the originating synchronisation quality is from a multiplexer SEC, SSU or Local PRC which meet the standards and is not degraded by the access E1, it will be suitable for synchronising a Node B. In addition, one of the new generation of GPS receivers is shown. If this were co-located with the Node B, it would also provide a suitable synchronisation reference. However this is probably too expensive to be cost effectively deployed at every Node B. We therefore need to examine other lower cost solutions for Node B timing.

Page 21: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 21 of 30 08 Feb. 06

Node B Sync Solutions Whilst we can develop theoretically ideal solutions for delivering an E1 with a quality fit for purpose at the Node B, it is not always an ideal world that we live in. Firstly any E1 delivered over SDH is at risk from VC12 pointers. These as we have seen can seriously damage the operational behaviour of a wireless base station. The wireless operator may not own his own SDH backhaul fibre. This resource may be provided by an independent 3rd party wireline carrier, or it may be over some free space medium which is not conducive to passing wander free sync. The 3rd party backhaul carrier may not be able or may not wish to provide access to the SDH line clock within his co-located access SDH mux. If access is allowed, there may be no guarantees for the sync quality. However the access SDH mux provides an excellent low wander and traceable path to good long term sync quality. We must look for a way to get this sync quality into the Node B. Similar issues exist with PDH. The wireless operator may not own his own backhaul copper or fibre. This resource may be provided by an independent 3rd party wireline carrier, or it may be over some free space medium which is not conducive to passing wander free sync. The 3rd party backhaul carrier may have carried the access PDH over a local access SDH ring. There is therefore a still a risk of SDH pointers being present on the access PDH E1. The access PDH must either be retimed to a known good clock at its launch point or some local retiming must take place in order to provide a firewall to VC12 pointers. If we are considering deploying telecom quality GPS at Node B’s – and this is vary different to timing or location quality GPS found in the low cost high street hand held GPS locators – we must select a solution which is quick and easy to fit whilst still being low cost and fit for purpose. Let’s examine four possible scenarios, two with Access PDH and two with Access SDH.

Page 22: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 22 of 30 08 Feb. 06

E1 Retimed to a Traceable Reference at Source The first option we will look at will be for an access PDH spur from the local SDH ring to the Node B. If the access PDH equipment collocated at the access SDH ring allows traceable retiming to a GPS/SSU appliance also co-located at that launch node for the PDH spur, this can be used to effectively separate the E1 from any VC12 pointers and buffer related wander which may be present. The resulting E1 delivered over the copper or fibre to the Node B will have an MTIE performance well inside the Node B MTIE and ideally well inside the ETSI PRC mask. Technically this latter is quite possible with low cost solutions available today.

SDHMultiplexer

Optional SSUor GPS

Receiver

TerminatingElement withE1 Retiming

Node BTerminatingElementE1

Freq DistributionAmplifier

Ethernet

Clean E1 retimed and pointer free

SDH Line Clock synchronisation reference

Optional local GPS or SSU reference

E1 from Mux, risk of pointers

Ethernet Management Communications

Node B Cabinet

Figure 14:E1 Retimed to a Traceable Reference at Source

Page 23: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 23 of 30 08 Feb. 06

Native E1 with no Retiming at Source The second PDH example is the more likely method which will be encountered in the real world. Here a PDH spur is provisioned from the access SDH equipment at the office nearest to the Node B. The E1 presented from the network terminating equipment co-located with the Node B will present native E1 clock at the Node B. This will show all the buffer wander associated with that E1 and there will be a risk of E1 pointers. The solution now is to retime the E1 at the Node B. Since there is no local source of low wander sync, this solution will need to look to GPS for this. Solutions now exist which can retime the E1 cost effectively to a low cost GPS engine and allow the Node B to “see” an E1 with wander well below the ETSI PRC mask. It is important within this architecture to consider another E1 provisioned directly to the Node B and not through the retiming element. This way if there is a failure of the retiming element, traffic backhaul will not be compromised.

SDHMultiplexer

Optional SSUor GPS

Receiver

TerminatingElement Node BTerminating

ElementE1

E1 Delivered over Copper or Fiber

Clean E1 retimed and pointer free

SDH Line Clock synchronisation reference

Optional local GPS or SSU reference

E1 from Mux, risk of pointers

Embedded SyncGPS

Node B Cabinet

Figure 15: Native E1 with no Retiming at Source

Page 24: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 24 of 30 08 Feb. 06

Access SDH – Retimed with SDH Line Clock If access SDH is employed as the communications link between the SDH ring and the Node B, things are a little easier. The line clock feed from the access SDH mux can be used as the external reference feed for the retiming element at the Node B. The only problem with this solution is that the 3rd party backhaul carrier may not allow or guarantee this connectivity to his local access SDH mux line clock. Again it is important within this architecture to consider another E1 provisioned directly to the Node B and not through the retiming element. This way if there is a failure of the retiming element, traffic backhaul will not be compromised.

Access SDHMultiplexer

Optional SSUor GPS

Receiver

Node B

Clean E1 retimed and pointer free

SDH Line Clock synchronisation reference

Optional local GPS or SSU reference

E1 from Mux, risk of pointers

Access SDHMultiplexer

EmbeddedSync

STM-N SDH Bearer

Node B Cabinet

Figure 16: Access SDH - Retiming with Line Clock

Page 25: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 25 of 30 08 Feb. 06

Access SDH Retimed with SDH Line Clock and Enhanced with GPS Even if a suitable Access SDH mux is supplied and connection to the line clock is allowed. It may still be prudent to utilise GPS as a fall back or as the first choice for retiming the E1. Again it is important within this architecture to consider another E1 provisioned directly to the Node B and not through the retiming element. This way if there is a failure of the retiming element, traffic backhaul will not be compromised.

Access SDHMultiplexer

Optional SSUor GPS

Receiver

Node B

Clean E1 retimed and pointer free

SDH Line Clock synchronisation reference

Optional local GPS or SSU reference

E1 from Mux, risk of pointers

Access SDHMultiplexer

Embedded Sync

STM-N SDH Bearer

Node B Cabinet

GPS

Figure 17: Access SDH – Retiming with Line Clock and GPS MTIE Quality at the Node B The terminating elements are the key to effective synchronisation transport to the Node B’s. The synchronisation quality is known at the local office SDH multiplexer, particularly if an SSU or GPS is present with MTIE measurement and reporting capability. PDH terminating elements which can transport E1’s, retime to a good quality synchronisation reference create the ideal Node B synchronisation model. SDH terminating elements offer traceability to low wander sync at the Node B which can then be used to retime the E1 at the Node B. There may be minor variations to

Page 26: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 26 of 30 08 Feb. 06

this model. For example, some SDH multiplexer manufacturers build a retimer into each E1 tributary. If traceability is not available a low cost GPS engine can be deployed as long as the MTIE is fit for purpose.

Figure 18: Quality of Sync available at the Node B using local synchronisation and retiming Figure 18 shows what can be achieved with latest technology equipment specifically designed with E1 retiming and better than PRC quality synchronisation transport capability. The egress E1 at the multiplexer end is retimed to either the multiplexer line clock or a local GPS/SSU. The E1 is passed over fibre and the MTIE shown is for the quality measured at the Node B end. Results are well within all masks, making this technique suitable for Node B timing. Conclusion This paper has really only scratched the surface of the issues relating to synchronisation in UMTS networks. The 3GPP standards are evolving continuously, their references are often circular. They are written without any regard for the risk imposed by significant wander which is always present in telecom networks due to buffer activity. Further work needs to be done to evaluate the Transport Channel Synchronisation, Radio Interface Synchronisation, Time Alignment Handling aspects of the UTRAN and what impact significant wander would have on the network performance.

Page 27: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 27 of 30 08 Feb. 06

This paper should provide a useful reference guide for UMTS network operators, manufacturers and backhaul bandwidth providers. The Node B MTIE proposed draws on information within the various 3GPP standards. The author welcomes any feedback relating to practical tests on UTRAN operation under stressed conditions which would help to fine tune the Node B MTIE mask and enable network designers to aim at a synchronisation quality necessary for efficient UTRAN operation. The issue of Location Services is just beginning to get in the radar of operators and users. New 3G networks will have to use radically different synchronisation techniques at the Node Bs if they are to offer a useable Location Service. It seems unlikely that the E1 method of Node B synchronisation will provide sufficient short and medium term stability to enable non-GPS based location services to work. Retiming of the E1 into the Node B from an access SDH STM-1 line clock (itself derived from GPS and Caesium) or provision of a local GPS at the Node B will provide suitable air interface stability. Some users for location services will no doubt include the emergency services. One question that will be asked is that of resilience and continuity of the Location Service. What will happen if GPS is temporarily switched off or reception is lost or degraded due to local interference? Then retiming to the access SDH line clock is the only viable alternative at the moment. There will be other off-air solutions in the future, but they are not available today.

Page 28: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 28 of 30 08 Feb. 06

Referenced Standards and Technical Documents. 3rd Generation Partnership Project/3GPP/ETSI TR 21.905: TR 121 905: “UMTS; Vocabulary for 3GPP Specifications” TS 03.71: TS 101 724: “Digital cellular telecommunications system (Phase 2+); Location Services (LCS); Functional description; Stage 2” TS 05.10: TS 100 912: “Digital cellular telecommunications system (Phase 2+); Radio subsystem synchronisation” TS 25.104: TS 125 104: “UMTS: UTRA (BS) FDD; Radio transmission and reception”. TS 25.105: TS 125 105: “UMTS: UTRA (BS) TDD; Radio transmission and reception”. TS 25.401: TS 125 401: “UMTS: UTRAN Overall Description”. TS 25.402: TS 125 402: “UMTS: Synchronisation in UTRAN Stage 2”. TS 25.411: TS 125 411: “UMTS: UTRAN Iu Interface layer 1” TS 25.421: TS 125 421: “UMTS: UTRAN Iur Interface Layer 1”. TS 25.431: TS 125 431: “UMTS: UTRAN Iub Interface Layer 1”. TS 45.010: TS 145 010: “Digital cellular telecommunications system (Phase 2+); Radio subsystem synchronization” ETSI EG 201 793: “Transmission and Multiplexing (TM); Synchronisation Network Engineering” EN 300 462-3-1: “Transmission and Multiplexing (TM); Generic requirements for synchronisation networks Part 3-1: The control of jitter and wander within synchronisation networks”. EN 300 462-4-1: “Transmission and Multiplexing (TM); Generic requirements for synchronisation networks Part 4-1: Timing characteristics of slave clocks suitable for synchronisation supply to Synchronisation Digital Hierarchy (SDH) and Plesiochronous Digital Hierarchy”. EN 300 462-5-1: “Transmission and Multiplexing (TM); Generic requirements for synchronisation networks Part 5-1: Timing characteristics of slave clocks suitable for operation in Synchronisation Digital Hierarchy (SDH) equipment”. EN 300 462-6-1: “Transmission and Multiplexing (TM); Generic requirements for synchronisation networks Part 6-1: Timing characteristics of Primary Reference Clocks” EN 300 462-7-1: “Transmission and Multiplexing (TM); Generic requirements for synchronisation networks Part 7-1: Timing characteristics of slave clocks suitable for synchronisation supply to equipment in local node applications". ITU ITU-T G.811: “Timing Characteristics of Primary Reference Clocks". ITU-T G.812: “Timing Requirements of Slave Clocks suitable for use as Node Clocks in Synchronisation Network". ITU-T G.813: “Timing Characteristics of SDH equipment slave clocks (SEC)".

Page 29: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 29 of 30 08 Feb. 06

ITU-T G.823: “The control of jitter and wander within digital networks which are based on the 2.048 kbit/s hierarchy” ITU-T G.824: “The control of jitter and wander within digital networks which are based on the 1.544 kbit/s hierarchy” ITU-T G.825: “The control of jitter and wander within digital networks which are based on the synchronous digital hierarchy (SDH)” Many of these standards and technical specifications are still undergoing revision, for the latest version numbers and further relevant standards browse the Chronos web site at: -

www.chronos.co.uk/pages/telecom/standards.html Glossary of Abbreviations and Acronyms 3G Third generation mobile wireless network - UMTS 3GPP Third Generation Partnership Project BS Base Station BTS Base Transceiver Station – Base Station for GSM networks Caesium Primary reference standard CN Core Network E1 2.048 Mbps digital telecom signal E-OTD Enhanced Observed Time Difference FDD Frequency Division Duplex GPS Global Positioning System – Satellite based navigation and

positioning system utilising high stability Caesium or Rubidium reference oscillators which are continuously corrected and traceable to UTC

GSM Global System for Mobile Communications Iu Interconnection point between an RNC and a Core Network.

It is also considered as a reference point. Iub Interface between an RNC and a Node B. Iur A logical interface between two RNC. Whilst logically

representing a point to point link between RNC, the physical realisation may not be a point to point link.

L1 Layer 1 LCS Location Services LMU Location Measurement Unit MS Mobile Station MTIE Maximum Time Interval Error NMC Network Management Centre Node B Base Station for 3G Networks PDH Plesiochronous Digital Hierarchy PRC Primary Reference Clock RF Radio Frequency

Page 30: UMTS Synchronisation

umts synchronisation.doc v1.31 Page 30 of 30 08 Feb. 06

RNC Radio Network Controller RNS Radio Network Subsystem SDH Synchronous Digital Hierarchy SEC SDH Equipment Clock SSU Synchronisation Supply Unit TDD Time Division Duplex TDEV Time Deviation UMTS Universal Mobile Telecommunications System – 3G UTC Universal Time Co-ordinated UTRA UMTS Terrestrial Radio Access UTRAN UMTS Terrestrial Radio Access Network VC12 SDH Pointer which may occur when mapping an E1 into the

SDH virtual container. Further relevant abbreviations can be found on the Chronos web site at: -

www.chronos.co.uk/pages/corp/abbreviation.html

The Author Charles Curry is Chairman of Chronos Technology Ltd. He has a Bachelor of Engineering (Hons) degree in Electronics from Liverpool University. His career started at GEC Hirst Research Centre where he was involved with research into semiconductor physics boundary properties. He then moved to Racal Instruments where he was responsible for sales of test equipment and frequency and time products. He moved into the rental business as sales manager with Micro Lease and then Managing Director of GSE Rentals in 1983. He founded Chronos Technology in 1986 and for the last ten years Chronos has been a leading system integrator of synchronisation products for the telecom industry. Charles created the International Telecom Sync Forum - ITSF which held its first 3 day seminar in London in 2001.

Document History Date Issue

October 2001 Preliminary Draft for Comments November 2001 Version 1.0 November 2001 Version 1.1 Inclusion of effect of wander

on Location Services over mobile networks

March 2002 Version 1.2 Details of speed of light and time errors added.

April 2002 Version 1.3 Various changes to incorporate generic Access and Node B sync solutions

February 2005 Version 1.31 – Correction of statement about VC12 pointers


Recommended