Joint ITU-T/IEEE Workshop on The Future of Ethernet Transport (Geneva, 28 ay 2010) Examples of Time Transport ichel Ouellette Technical Advisor Huawei Technologies Co., Ltd. Geneva, 28 ay 2010
Outline LTE-Advanced Time ynchronization Examples Time Distribution over Packet Transport Network Latest Field Trials Performance Results & Voice/Data ervice Results ummary 2
LTE-Advanced Network ynchronization Overview LTE witching Centers Network Based ynchronization Error (frequency and/or time) 1 Backhaul Network Over-the-Air ynchronization Error U N Relay Node U U enb X2 UE 3GPP TR36.814 defines further advancements for LTE LTE / LTE-A phase/time specs are beyond current G/UT frequency specs (50ppb) imposing further requirements on the transport network 3
LTE-Advanced Coordinated ulti-point Transmission* Radio transmissions from multiple base stations radiate at the same time (spatial multiplexing), giving better cell edge performance and higher data rates Broadcast service on the air interface Accurate time synchronization required at base station (GP, IEEE1588-time, etc.) for aligning radio transmissions Backhaul Network LTE witching Centers 1 t 2 0 Time difference = max{ t 0 -( 1 +t 1 ), t 0 -( 2 +t 2 ) } x µsec enb t 1 t 2 *Illustrative example Time Difference 4
LTE-Advanced Relay Function* Relay used to improve the coverage of high data rates sites, temporary network deployment, coverage in new areas, etc. The relay node (RN) is wirelessly connected to radio-access network via a donor cell (enb). The UE may need to get timing info from RN over the air, and where the RN also obtains timing over the air from the donor enb. ome relay nodes have strigent time alignment such as Example 2 RN nodes might be equipped with GP or is synchronized via the radio interface Donor enb might be equipped with GP or is synchronized via IEEE1588-time donor enb UE donor enb UE t 2 t 1 1 Example 1 t 2 t 1 1 Example 2 RN RN *Illustrative example 5
Field Trial Topology City-Wide Time Distribution, Joint with China obile Working BIT (Time ource) Protected BIT (Time ource) 1PP + ToD Ethernet BCA aster-lave hierarchy (time reference chain) Boundary clocks in each device BC BC GE 1PP+ToD GE 1PP+ToD BC 10GE Core Ring (Packet Transport Network with IEEE1588) P GE Access Rings with IEEE1588 P NodeB NodeB FE 1PP+ToD 1PP+ToD Requirement at NodeB: < ± 1.5usec of cumulative time error FE 1PP+ToD (output) 6
Field Trial etup Time Distribution One Core 10GE ring with 6 Access GE ring Approximately 75 nodes, covering one city, multi-vendor equipment Two Time Reference ources Dual mode GP/Beidou receivers, connected to packet transport nodes via Gigabit Ethernet interface (running IEEE1588) or direct 1PP+ToD interface Packet Transport Nodes IEEE1588 Boundary Clock & Best-aster Clock Algorithm. IEEE1588 terminated in each port, 10GE & GE timestamping (similar in spirit to 802.3bf architecture) ynchronous Ethernet & EC channel for frequency distribution IEEE1588 time and ynce frequency reference chain are congruent Fiber asymmetry compensated section-by-section NodeB base stations Receive time synchronization from Fast Ethernet interface (running IEEE1588) or direct 1PP+ToD coaxial input. NodeB can output 1PP interface for measurement purposes Traffic PTN carries real voice/data services easurement equipment TimeAcc 1PP+ToD analyzer, call quality drive test tool and voice call generator Test cenarios Long term cumulative time error performance Time Reference protection switch, PTN protection switch BCA master-slave hierarchy, ynce reference switch Quality of voice/data service, handoff and call completion 7
Performance Results Cumulative Time Error (nsec) versus Time at NodeB 40 nsec 35 30 25 20 15 cenario1: BIT uses 1PP+ToD to provide time to packet transport, and BT uses IEEE1588 FE to recover time 10 5 0-5 1 29 57 85 113 141 169 197 225 253 281 309 337 365 393 421 449 477 505 533 561 589 Result: Acceptable performance nsec 145 140 135 130 125 120 115 110 105 1 33 65 97 129 161 193 225 257 289 321 353 385 417 449 481 513 545 577 609 cenario2: BIT uses 1PP+ToD to provide time to packet transport, and BT uses 1PP+ToD interface to recover time Results: NodeB not capable of compensating (at the time of the trial) for 1PP+ToD coaxial cable delay and internal delay Delay compensation of every single element is necessary 8
Performance Results Long term Cumulative Time Error (36 hours) nsec 0 12 hours nsec 12-24 hours Long-term cumulative time error < ± 1.5usec nsec 24-36 hours 9
Voice/Data Quality Results Drive tests Between NodeBs supporting IEEE1588 time sync Between NodeBs supporting IEEE1588 and NodeBs supporting GP Voice/Data ervices under test & Handoff performance AR voice, VP and P384 (various wireless codecs) Eg., AR traffic: 100-seconds call with 10-seconds interval Drive speed: 40 60km/hour Tests (reference C1065 Geneva June 2010) uccessful handoff ratio between adjacent base stations: 100% Call completion ratio: 100% Average voice O (mean opinion score): 3.46 Drive test routes 10
ummary Accurate Phase/Time distribution over large-scale Packet Transport Network is progressing within Q13, primarily serving wireless radio interfaces requirements Various Field Trials IEEE1588 distribution over Packet Transport Network IEEE1588 distribution over Optical Transport Network Field trials using Boundary Clocks & Best aster Clock show good results Cumulative time error < ±1.5 usec for TD-CDA NodeB Packet impairments: reference switch, protection switch Good voice & data quality of service uccessful handoff and call completion Additional references: C599, ZTE/CCC, Geneva, October 2009 WD29, Huawei/CCC, Lannion, December 2009 C1065, ZTE/Huawei/CCC/CATR, Geneva, June 2010 C1064/65, ZTE/CCC/CATR, Geneva, June 2010 11
Additional lides Additional performance results Time distribution via IEEE1588 Frequency distribution ynchronous Ethernet Boundary Clock 12
Trial Results GP GP 2 time synchronization reference chains BIT1 BIT2 A chain consists of 15 nodes Each node implements IEEE1588 (for time) and ynchronous Ethernet (for frequency) Default Best aster Clock Algorithm used to establish aster-lave hierarchy and electing new Grandaster during failure Chain #1 52 km G1 1PP+ToD 10GE RAN P 1PP+ToD G2 Chain #2 43km G1 is the initial Grandaster of the network 1PP signal (cumulative time error) and 2.048Hz signal (time interval error) used for time and frequency measurements GE RAN P : aster port : lave port (only 1 per switch) P: Passive port GP reference Time Tester 1PP Fiber G: Grandaster 13
Frequency and Time Results Performance of two Boundary Clocks connected via single GE link ± 3 nsec on 2.048Hz output Frequency (ynce) ~= 10-11 frequency offset PRC traceable results Time (IEEE1588) ± 2 nsec 1PP pk-to-pk jitter + 3 nsec static time error due in part to fiber asymmetry 14
Frequency and Time Results Protection scenario from Chain#1 to Chain#2 aster failure done by disconnecting 1PP interface/cable from G1 aster2 is the new Grandaster, produceed by BCA & the exchange of Announce messages. Time is distributed through Chain #2 + 80nsec tatic Time Error (eg., fiber asymmetry, timestamping error, temperature) Chain #1 Time error performance (± 15 nsec) Protection witch from Chain 1 2 Phase Transient (~ 60 nsec) Chain #2 Time error performance (± 20 nsec) Results demonstrate the robustness and performance when using IEEE1588 node-by-node time synchronization 15
Boundary Clock ystem Ext ync Input Timing Card ynce EEC ystem Clock ynce RX clock 1PP + local time timestamp, id (T2, T3) 1588 Clock 1588 oftware (T1, T2, T3, T4) BCA, etc. 1PP + local time PHY PHY Timestamp Capture (T2, T3) AC witch Fabric AC Timestamp Capture (T1, T4) PHY PHY Line Card (1588 lave Port) Line Card (1588 aster Port) Boundary Clock is similar in concept to current DH/ONET/yncE synchronization system (deal with time and not just frequency) 16