Performance Comparison Based on SMG 2 Evaluation Reports: WCDMA vs. WB-TDMA/CDMA

Transcription

1 ETSI SMG #24 Tdoc SMG 99 / 97 Madrid, Spain December, 5-9, 997 Source: Ericsson Performance Comparison Based on SMG 2 Evaluation Reports: WCDMA vs. WB-TDMA/CDMA Contents. INTRODUCTION LINK LEVEL RESULTS...2. LINK BUDGETS - COVERAGE SYSTEM LEVEL RESULTS - CAPACITY REFERENCES...7 Executive Summary: This paper contains the WCDMA and WB-TDMA/CDMA performance results as they are presented in the ETSI evaluation reports [][2]. Results are presented for link level simulations (E b /N 0 ), link budget (coverage) and system capacity. Since the set of results are different for the two concepts only the results for the prioritised test cases are shown, as well as those cases where both concepts have submitted results to SMG2. In addition to the results, the assumptions used to obtain them are also described. When comparing the different results for the system capacity, it is shown that the WCDMA concept gives better results in all cases and especially in low-speed environments. It should be noted that there is a difference in how transmission overhead, e.g. common control channels, are taken into account. It has been taken into account for WCDMA but not for WB-TDMA/CDMA when determining the capacity results. Also effects from adjacent channel interference has not been taken into account for any of the concepts. However the impact of adjacent channel interference on the capacity differs between the concepts. It has a larger impact on WB-TDMA/CDMA performance than on WCDMA performance. This shows that the tendency of capacity results in favour of WCDMA will be even larger. Finally, it should once again be emphasised that all results in this document are directly taken from the SMG2 evaluation reports and related documentation. /7

2 . Introduction This paper contains a summary of link level simulation results (E b /N 0 ), link budgets (coverage), and system level simulation results (capacity) provided for WCDMA and WB-TDMA/CDMA by the α- and the δ-concept groups during the evaluation phase in SMG2. All results presented here are directly taken from the evaluation reports of the α- and δ-group submitted to SMG #24 ([] and [2] respectively) and other contributions to SMG2. The presented test cases are the test cases defined by SMG2 in Tdoc SMG2 260/97 and 29/97. In addition, some cases not prioritised by SMG2, but where both concept groups have submitted results to SMG2 are also presented. However, no UDD link level results are presented since these are not directly comparable (see Link Level Results ). 2. Link Level Results In Table the link level results provided by the α- and the δ-concept groups are summarised. The WCDMA results are taken from [], part, Section.2., whereas the WB-TDMA/CDMA results are taken from [2], part, Section.. Since the E b /N 0 requirements for WB-TDMA/CDMA increase with the number of users per time slot (K) [2, part, Section.], there are several possible values available for each test case. In this paper the lowest and highest K-values that are presented in [2] are shown. For instance, speech values are presented for K=,4 and in [2]. The K-values and are then used in Table. It is not clear how many codes per time slot WB-TDMA/CDMA is supposed to handle with respect to terminal complexity etc. However, in the system level simulations K can be as high as 2, but since link level results have not been presented in the WB-TDMA/CDMA evaluation report no figures can be shown for this case. The WCDMA values presented are the same as the ones used for deriving the WCDMA link budgets and capacity results. This is not the case for the WB-TDMA/CDMA values. The WB-TDMA/CDMA link budgets are derived assuming a single user per time slot (K=), since a low load is assumed in the link budgets [2, part and 5 together]. The WB-TDMA/CDMA link level results used in the system level simulations depends on the system load (i.e. K = -2) [2, part 4]. Further, for WCDMA, rate information (RI) has been used for all results, except for the speech cases [, part, Section.2]. For the WB-TDMA/CDMA results, a minimum mean square error (MMSE) equaliser in WB- TDMA/CDMA has been used in the simulations [2, part, Section.], instead of the previously used zero-forcing equaliser. Also, a receiver filter matched to the linearized GMSK pulse has been introduced [2, part, Section.]. Finally, there are no UDD results presented in Table. This is due to the fact that the α-group has presented E b /N 0 figures at 0% block error rate [, part, Section.2], whereas the δ-group has presented C/I at a throughput equal to the long term average source bit rate (of the packet model). It is therefore believed that the UDD link-level results cannot be compared [2, part, Section..]. Rate Information, RI, is information to the receiver of what bit rate that is currently used. 2/7

3 Table. Summary of the Link Level Results. WCDMA values are taken from [], part, Section.2. and WB-TDMA/CDMA values are taken from [2], part, Section.. Link Level Results WCDMA E b /N 0 [db] Environment Service Vehicular A, 20 km/h Vehicular B, 20 km/h Vehicular B, 250 km/h Outdoor Pedestrian A, km/h Indoor Office A, km/h Speech, - LCD 4 Speech, - Speech, - Speech, - LCD 44 Speech, - LCD 4 LCD 204 WB-TDMA/CDMA E b /N 0 [db] Ideal FH No FH 5.0 / / - 5. /.7. / / - (a) 5.0 / 9..6 / 9. (b) 4.9 / / - 5. / / / / 0.. / / / 0.2. / 4..7 / - (a) 4. / / 9. (b) 2.2 / - (c) 2.6 / 7.6. / / / / / - (a).9 / / 7.4 (b) 0. / / -.4 / - # users per time slot (K).0 / / 0.9 Explanations: BER = Bit Error Rate FH = Frequency Hopping Vehicular A: Max. relative delay = 2.5 µs. Vehicular B: Max. relative delay = 20 µs., i.e., more time dispersion than in vehicular A. K= is the only choice for LCD204, LCD44 b) and LCD4 b). LCD44 a): time slots per user, code per time slot LCD44 b): time slot per user, 9 codes per time slot LCD44 c): 4 time slots per user, codes per time slot LCD4 a): time slots per user, codes per time slot LCD4 b): time slots per user, 9 codes per time slot. As seen from the table, the link-level results for the two proposals are very similar for the Vehicular environment (within ±0.5 db). On the other hand, for the indoor and pedestrian environment WCDMA has an advantage in the range db. The main reason for this difference is the use of fast closed-loop power control in WCDMA which very effectively combats the fading. For the vehicular environment, there is already a significant amount of diversity against fading due to the fast terminal speed and time dispersion. Consequently the two concepts have similar performance in this environment. /7

4 . Link budgets - Coverage In Table 2 the link budgets provided by the α- and the δ-concept groups are summarised. The WCDMA values are taken from [], part 2, Section, whereas the WB-TDMA/CDMA values are taken from [2], part 5, Section 2.4. Please note that the link budgets for WB-TDMA/CDMA are only valid for user per time slot (K=). This is because the link budget results are based on the E b /N 0 values for K= [2, part and 5 together] as the link budgets should reflect the low-load scenario. For a loaded system, a higher value of K should be used with a corresponding degradation in the link budget. The maximal path loss of WB-TDMA/CDMA thus depends on the load. The same is true for WCDMA. No WB-TDMA/CDMA link budgets and range estimates are available for the UDD services. Table 2. Summary of the Link Budget Results. WCDMA values are taken from [], part 2, Section and WB-TDMA/CDMA values are taken from [2], part 5, Section 2.4. Environment LINK WCDMA WB-TDMA/CDMA BUDGETS Ideal FH K= No FH K= Service Maximal Path Loss [db], UL Maximal Path Loss [db], UL Maximal Path Loss [db], UL Vehicular A 20 km/h Speech, kbps LCD 4 kbps (a) 7.57 (b) Outdoor Pedestrian A Speech, kbps km/h LCD 44 kbps. 2.5 (a) 24. (a) 2.24 (b) 0. (c) UDD UDD Indoor Office A Speech, kbps km/h LCD 4 kbps (a) (b) LCD 204 kbps UDD UDD Explanations: K = Number of users per time slot. K= is the only choice for LCD204, LCD44 b) and LCD4 b). FH = Frequency Hopping LCD44 a): time slots per user, code per time slot LCD44 c): 4 time slots per user, codes per time slot LCD4 a): time slots per user, codes per time slot LCD4 b): time slots per user, 9 codes per time slot WCDMA UDD results are for 0% block error rate (BLER) The link budgets are very much in accordance with the link-level results of Table. For the Vehicular environment, the maximum path loss of the two concepts are very similar. On the other hand for the 4/7

5 indoor and pedestrian environments, there is an advantage for WCDMA of up to db. For the case when WB-TDMA/CDMA does not use frequency hopping, e.g. due to limited spectrum allocation, the loss is larger (up to 9 db). The reason for the better link budget of WCDMA is partly the better link-level figures of Table but also the additional handover gain due to soft handover in WCDMA. 4. System Level Results - Capacity In Table the system simulation results are summarised. The WCDMA results are taken from [], part, Section.2.2, whereas the WB-TDMA/CDMA results are taken from [2], part 4. When it comes to system level simulations there should be some words mentioned about the models and assumptions. However, here we only point at some differences in the assumptions, directly taken from the material, provided by the concept groups. All WB-TDMA/CDMA results are produced while assuming a maximum of 2 codes per time slot [2, part 4]. Further, The DTX is handled differently: In WCDMA the layer control signalling is not turned off during silent speech periods [, part, Section.2.2.]. However, in WB-TDMA/CDMA, all transmissions are stopped during the silent speech periods, which implies that no signalling is allowed during the silent periods ([2], part 4). The amount of signalling included differs as well: Both concepts has included layer signalling (WCDMA: power control commands, rate information and pilot; WB-TDMA/CDMA: midamble), but WCDMA has also included all downlink common control channels in the capacity simulations. This has not been considered for WB-TDMA/CDMA [7]. Further, around 5 MHz bandwidth has been assumed in the simulations. For WCDMA a carrier spacing of 5 MHz has been assumed [2, part, Section.2.2], thus the guard bands are included [6]. In WB-TDMA/CDMA, the guard bands have not been considered [2, part 4]. There is no adjacent channel interference assumed in the simulations. This will according to the Delta concept group decrease the WB-TDMA/CDMA capacity by 25% in the case of re-use and by 0% in the case of a clustered system [5, Section 2.] For WCDMA the capacity loss would be less than % with a carrier spacing of 4.4 MHz [6, Section 2]. 5/7

6 Table. Summary of the System Level Results. WCDMA results are taken from [], part, Section.2.2 and WB-TDMA/CDMA results are taken from [2], part 4. System Level Results WCDMA WB-TDMA/CDMA - Capacity - Environment Service Note Vehicular A, 20 km/h Speech, kbps LCD 4 kbps Noise limited Interf. limited No DL Ant. DL Ant. No DL Ant. Spectrum efficiency [kb/s/mhz/cell] 9 / 7 NoFH: - / 6 Spectrum efficiency [kb/s/mhz/cell] FH: - / 72 / 5 No reasonable capacity figures can be provided [2, part 4] / 2 No reasonable capacity figures can be provided [2, part 4] Re-use 204 / 2 - / 29 (a) - - DL Ant. 204 / / 76 (a) Outdoor Pedestrian A, km/h Indoor Office A, km/h Speech, kbps UDD 4 UDD 204 No DL Antenna 27 / 6 NoFH: - / 7 FH: -/ / /- - DL Ant. - / 7 -/ 2 No DL Antenna 00 / 20 - / 2 - / 70 DL Ant. 00 / / 95 - / 405 Mixed: Speech & No DL Antenna 5 / UDD 4 UDD DL Ant. 5 / 460 NoFH - / 04 FH: - / 0 Explanations: FH = Frequency Hopping LCD4 a): time slots per user, codes per time slot 6/7

7 5. References [] Tdoc SMG 905 / 97, Concept group Alpha - Wideband Direct-Sequence CDMA (WCDMA), Evaluation Document (.0). [2] Tdoc SMG 99 / 97, Concept group Delta, WB-TDMA/CDMA, System Description Performance Evaluation [] ETSI TR 0 2 (UMTS 0.0) Selection procedure for the choice of radio transmission technologies of the UMTS [4] SMG2 UMTS ETR 04-0, version 2.6., Requirements for the UMTS Terrestrial Radio Access System (UTRA). [5] Tdoc SMG2 UMTS 29 / 97 Answers to Questions by Ericsson to the Delta group, Evaluation Report, part 4 [6] Tdoc SMG2 2 / 97 Guard bands for WCDMA. [7] ETSI SMG2 UMTS Ad Hoc meeting, Helsinki, Finland, November 7-2, 997 7/7