Simulation-Base Performance Evaluation in and -Advanced João Gonçalves, n.º 57940 Instituto Superior Técnico Universidade Técnica de Lisboa Av. Rovisco Pais, 1049-001 Lisbon, Portugal joao.goncalves@ist.utl.pt Abstract The main objective of this paper is to simulate the throughput of the new generation of mobile networks and check if the results are consistent with theoretical values. The network and its evolution, -Advanced, were studied. Using two different simulators, the parameters that had a greater impact on maximum throughput were studied and the conclusion was that these parameters are the Bandwidth, the value of Channel Quality Indicator (CQI) and the antenna configuration (through MIMO techniques). For this reason, the impact of the variation of each one of these parameters in the maximum throughput was studied. The conclusion is that the higher the bandwidth, the higher the maximum throughput, however, the spectral efficiency increases only slightly. It was also found that with the increase of the CQI value, the maximum throughput and spectral efficiency greatly increase and the same can be verified when increasing the number of transmitting or receiving antennas. This occurred for both cases ( simulator and -Advanced simulator), but it was also found that the simulator originated values closer to theoretical values while the -Advanced simulator simulates better the conditions that will be found in reality. Through the two simulators it was possible to verify much of what was stated in the initial chapters of the thesis, allowing the main objective to be achieved. Keywords:, -Advanced, Bandwidth, CQI, MIMO I. Introduction The 3GPP was proposed and presented in Release 8 [1] and 9 [2] to evaluate a new whole system for Mobile Communications. The new architecture reflects the uptake of IP-based services in mobile communications, likewise the full optimization of network performance and improve costefficiency. network architecture has several main differences from the UMTS one, being a flat architecture that reduces the involved nodes in the connections, besides a hierarchical one. With this new architecture it is possible to reduce the latency to 10 ms, and with use of new technologies (for example, MIMO) it is possible to generate high data rate (around 100 Mbps for DL and 50 Mbps for UL). However, it does not meet certain requirements established by the International Mobile Telecommunication-Advanced (IMT-Advanced) [3], and therefore cannot be called as 4G network. The IMT- Advanced is a concept for mobile systems with capabilities beyond IMT-2000 [4] and therefore was previously known as System beyond IMT-2000. Throughout 2009 there was a 1 period for submitting applications to the IMT-Advanced to be submitted to ITU-R, as well as the start of assessment activities of candidate technologies and systems. The deadline for sending the radio interface for the submissions was October 2009 and the deadline for final specifications was 2011, [5]. The new capabilities of IMT-Advanced systems are planned to deal with a wide range of supported data rate according to economic and service demands in multi-user environments with target peak data rate for up to 100 Mbps (high mobility) and up to 1 Gbps (low mobility or static). The 3GPP has also started to work for the goals of IMT- Advanced to the local area radio under the name - Advanced. -Advanced is part of Release 10 of the 3GPP and the commercial deployment of IMT-Advanced should be in 2013 or later. The high level evolution of 3GPP technologies in order to meet the requirements of IMT is shown in Figure I.1. Figure I.1 Bit rate and mobility evolution to IMT-Advanced. [5] Enhancements relative to Releases 8 and 9, [6], [7], [8]: Carrier Aggregation (CA): o Contiguous and non-contiguous; o Control channel design for UL/DL. Enhanced DL multiple access schemes; Enhanced UL multiple access scheme (Clustered SC- FDMA); Enhanced DL MIMO transmission; Enhanced UL MIMO transmission. MIMO systems promise to deliver large data rate to users, e.g. several tens of megabits per second, which is substantially larger than what the current 3G systems offer. This is a challenging proposition as wireless networks are subject to interference, multipath and poor propagation channel characteristics that limits data rates. MIMO techniques have emerged as a solution to provide higher data rates by
exploiting the multipath characteristics of the wireless channel. This is accomplished by using several antennas to transmitting and receiving signals thereby leveraging the spatial dimension resulting from using multiple spatially distributed antennas (hence the term multiple input multiple output: MIMO). When the signals are combined properly at the receiver, the signal quality or data rate for each MIMO user will be improved. Figure I.2 Multiple Antenna Access Schemes. [9] A key factor to the performance of MIMO is the number of spatial layers of the wireless channel which determines the ability to improve spectral efficiency. Spatial layers are born out of the multipath and scattering environment between transmitters and receivers. Another factor is the number of transmitting and receiving antennas. The increase in data rate of a MIMO system is linearly proportional with the minimum number of transmitting and receiving antennas subject to the limit of the rank of the propagation channel estimate. The rank is a measure of the number of independent spatial layers. Hence, a 4TX/2RX antenna MIMO system provides double the data rate (i.e., min(4,2) = 2) provided that there are two spatial layers (rank = 2) in the wireless channel. In line-of-site conditions, the channel matrix rank is one; hence, even with 4 antennas we cannot increase the spectral efficiency of the channel, [9], [10]. CQI In cellular communication systems, the quality of the signal received by a UE depends on the channel quality from the serving cell, the level of interference from other cells, and the noise level. To optimize system capacity and coverage for a given transmission power, the transmitter should try to match the information data rate for each user to the variations in received signal quality (see, for example, [11] and [12] and references therein). This is commonly referred to as link adaptation and is typically based on Adaptive Modulation and Coding (AMC) [13]. The degrees of freedom for the AMC consist of the modulation and coding schemes: Modulation Scheme Low-order modulation (i.e. few data bits per modulated symbol, e.g. QPSK) is more robust and can tolerate higher levels of interference but provides a lower transmission bit rate. High-order modulation (i.e. more bits per modulated symbol, e.g. 64AQM) offers a higher bit rate but is more prone to errors due to its higher sensitivity to interference, noise and channel estimation errors; it is therefore useful only when the SINR is sufficiently high. Code rate For a given modulation, the code rate can be chosen depending on the radio link conditions: a lower code rate can be used in poor channel conditions and a higher code rate in the case of high SINR. The adaptation of the code rate is achieved by applying puncturing or repetition to the output of a mother code. In the UE can be configured to report CQIs to assist the enodeb in selecting an appropriate MCS to use for the downlink transmissions. The CQI reports are derived from the downlink received signal quality, typically based on measurements of the downlink reference signals (see Section 8.2). It is important to note that, like HSDPA, the reported CQI is not a direct indication of SINR in. Instead, the UE reports the highest MCS that it can decode with a transport block error rate probability not exceeding 10%. Thus the information received by the enodeb takes into account the characteristics of the UE s receiver, and not just the prevailing radio channel quality. Hence a UE that is designed with advanced signal processing algorithms (for example, using interference cancellation techniques) can report a higher channel quality and, depending on the characteristics of the enodeb s scheduler, can receive a higher data rate. However, a trade-off exists between the amount of CQI information reported by the UEs and the accuracy with which the AMC can match the prevailing conditions. Frequent reporting of the CQI in the time domain allows better matching to the channel and interference variations, while fine resolution in the frequency domain allows better exploitation of frequency-domain scheduling. However, both lead to increased feedback overhead in the uplink. Therefore, the enodeb can configure both the time-domain update rate and the frequency-domain resolution of the CQI, [13]. The list of modulation schemes and code rates which can be signaled by means of a CQI value is shown in Table I.1. Table I.1 CQI Table. [13] CQI Modulation Code Rate Bits/Symbol 0 Out of range --- --- 1 QPSK 0,076 0,1523 2 QPSK 0,120 0,2344 3 QPSK 0,190 0,3770 4 QPSK 0,300 0,6016 5 QPSK 0,440 0,8770 6 QPSK 0,590 1,1758 7 16QAM 0,370 1,4766 8 16QAM 0,480 1,9141 9 16QAM 0,600 2,4063 10 64QAM 0,450 2,7305 11 64QAM 0,550 3,3223 12 64QAM 0,650 3,9023 13 64QAM 0,750 4,5234 14 64QAM 0,850 5,1152 15 64QAM 0,930 5,5547 2
II. Description of the Simulators To perform the simulations on, it was used the Link Level simulator[14], v1.6r917, developed by the Institute of Communications and Radio-Frequency Engineering of the Vienna University of Technology. This simulator, developed in Matlab [15], consists of dozens of scripts that reproduce many of the features of. The idea was to run simulations so that the output was the throughput as a function of the SNR and analyse the spectral efficiency of each simulations after. Through the first simulation it was verified that this simulator simulates conditions very close to ideal conditions, since the throughput values generated are very close to theoretical values. To perform the simulations of -Advanced, it was used the SystemVue 2011.03 simulator [16], developed by Agilent Technologies [17], which was launched in April 2011. This simulator consists on several libraries, but for the simulations performed, only the W1918 -Advanced Baseband Verification [18] library was used. Since this was a paid software, a provisional license for student was necessary however, in total, it gave access to the software for only 45 days. This simulator, unlike the simulator used in, didn t consists of scripts, but it was composed of schemes that were design to simulate the network. Then analyses and sweeps were ran in order to calculate the throughput generated as a function of SNR. It was also found that, unlike simulator, this simulator was much more sensitive to small variations so, the graphics drawn oscillate considerably for each value of SNR. The receptor weren t also fully configured properly, and for this reason, the values were lower than expected taken. However, it was possible to verify that this simulator simulates conditions very close to reality. Based on the use of these two simulators was possible to verify that the parameters that had a greater impact on throughput generated were Bandwidth, CQI value and antennas configuration (using MIMO techniques). Were also performed some simulations where other parameters were varied, such as the channel type, the number of subframes, or the use of retransmissions, but quickly became apparent that these parameters had no major impact on throughput generated. To point out that, for the various simulations were used for the constant parameters (those who aren t in analysis) the values presented in Table II.1: Table II.1 Constant parameters in the several simulations. Number of Channel CQI Value Bandwidth Subframes Type 500 PedA 7 5 MHz MIMO Mode Number of Transmitting Number of Receiving 3 2 2 III. Results Analysis In order to simulate de impact of the variation of Bandwidth in, a script where this parameter was varied keeping other parameters constant was developed. In the case of -Advanced simulator a scheme was design and the 3 bandwidth was also varied. Based on these two simulations, it was possible to reach the results presented in the following tables (Table III.1 e Table III.2). Table III.1 Values of throughput and spectral efficiency for each of the existing bandwidth in. Bandwidth [MHz] Maximum 1,4 2,557 1,826 3 6,602 2,201 5 11,080 2,216 10 22,310 2,231 15 33,530 2,235 20 44,750 2,238 Table III.2 Values of throughput and spectral efficiency for each of the existing bandwidth in -Advanced. -Advanced Bandwidth [MHz] Maximum 1,4 1,014 0,724 3 2,783 0,928 5 4,625 0,925 10 9,452 0,945 15 14,555 0,970 20 18,930 0,946 Analyzing the Table III.1 and Table III.2 it is possible to see that, as expected, the throughput increases dramatically with the increase in bandwidth. However, it also transpires that the spectral efficiency is similar for each of the bandwidth, in each of the simulators. It s possible to see that, also in both simulators, there is a slight increase in spectral efficiency with the increase of bandwidth, although this is most evident in the simulator than in the - Advanced simulator. This is because the -Advanced simulator is much more sensitive, oscillating with small variations. The ratio between the value of the throughput in the -Advanced simulator and the simulator is more a less 40%. So, the values in the -Advanced simulator are much smaller than the ones in the simulator. This different was expected, as noted above, since the receptor in the -Advanced simulator are not yet fully configured properly and the values of the simulator are very close to the theoretical values. Then, the impact of the variation of CQI value in the throughput generated was simulated. In the case if simulator, a script where the CQI value was varied between 1 and 15 (CQI values specified in and -Advanced) was developed. To do this simulation in the -Advanced simulator, is was necessary to convert the CQI value in modulation and code rate, since the CQI parameter is not defined in the Agilent Technologies simulator. To do that, it was necessary to use the Table I.1, where it is possible to check the modulation and code rate for each CQI value. Based on these two simulations, it was possible to build the following tables (Table III.3 e Table III.4).
Table III.3 Values of throughput and spectral efficiency for each of the existing CQI values in. CQI Value Maximum 1 1,098 0,220 2 1,715 0,343 3 2,794 0,559 4 4,493 0,899 5 6,573 1,315 6 8,813 1,763 7 11,080 2,216 8 14,390 2,878 9 18,100 3,620 10 20,550 4,110 11 25,010 5,002 12 29,390 5,878 13 34,070 6,814 14 38,540 7,708 15 41,840 8,368 Table III.4 Values of throughput and spectral efficiency for each of the existing CQI values in -Advanced. -Advanced CQI Value Maximum 1 0,648 0,130 2 0,777 0,155 3 1,290 0,258 4 2,034 0,407 5 2,870 0,574 6 3,724 0,745 7 4,628 0,926 8 6,019 1,204 9 7,549 1,510 10 8,478 1,696 11 10,699 2,140 12 12,638 2,528 13 14,419 2,884 14 17,183 3,437 15 17,487 3,497 Looking at Table III.3 and Table III.4 it appears that by increasing the CQI value the maximum throughput is greatly increased in the two simulators and, of course, the spectral efficiency is also greatly increased. This result was expected, since, with the increasing of CQI value, the modulation and code rate increase and, as such, the number of bits per symbol increases. If you re-calculate the ratio between the value of the throughput in the -Advanced simulator and in the simulator you can verify that it s again around 40%. Again, this result was already expected, as mentioned earlier. Finally, the impact of the antennas configuration in the throughput generated was simulated. In simulator, a script was developed where the number of transmitting and receiving antennas was varied. In the -Advanced simulator it was necessary to design schemes for SISO 1x1 and MIMO 4x4, because such schemes weren t designed in this simulator. Note that in both simulators the MIMO mode 3 was used, and in -Advanced simulator it wasn t 4 possible to simulate all the antennas configurations that were simulated in simulator. Based on these two simulations, it was possible to build the following tables (Table III.5 e Table III.6). Table III.5 Values of throughput and spectral efficiency for each of the existing Configuration in. Configuration Maximum SISO 1x1 5,838 1,168 MISO 2x1 5,542 1,108 MIMO 2x2 11,080 2,216 MISO 4x1 5,246 1,049 MIMO 4x2 10,490 2,098 MIMO 4x4 20,990 4,198 MIMO 8x8 --- --- Table III.6 Values of throughput and spectral efficiency for each of the existing Configuration in -Advanced. -Advanced Configuration Maximum SISO 1x1 2,419 0,484 MISO 2x1 --- --- MIMO 2x2 4,632 0,926 MISO 4x1 --- --- MIMO 4x2 --- --- MIMO 4x4 8,206 1,641 MIMO 8x8 17,137 3,427 Through the analysis of Table III.5 and Table III.6 it s possible to see that, in both simulators, the increase in the number of antennas also increases the maximum throughput and, of course, the spectral efficiency. It appears that this increase is proportional to the minimum number of antennas, as stated by the function min(ntx;nrx). Again, calculating the ratio between the value of the throughput in the -Advanced simulator and in the simulator it s possible to verify, by comparing the same configuration in both simulators, that it reaches around 40%. IV. Conclusions and Future Work This is a turning point for Mobile Communications. This new generation completely changes the paradigm of Mobile Networks. With a flatter architecture, with fewer nodes between the user and the network core, it s possible to reduce responses time for a few milliseconds. This is important for users since it closer and closer the mobile networks from fixed networks in terms of response times. It is also very important for some user s applications (e.g. games) because by 10 ms you can win or lose and the mobile operators will enjoy these benefits of this new generation to bet on applications to meet the user s requirements. With the news technologies implemented is also possible to increase the data rate of this networks, reaching up to 100 Mbps (DL) and 50 Mbps (UL) (in ), something unthinkable in 3G networks. In terms of simulations, it was possible to verify that only a few parameters have a significant influence on the
throughput generated. This is the case of bandwidth, CQI value and antennas configuration (using MIMO techniques). For the first case (bandwidth), it s possible to use six different bandwidths, from 1,4 MHz up to 20 MHz (flexible bandwidth), depending on user s needs. The higher the bandwidth, the greater the number of resource blocks allocated to each user. It was also found that the higher the bandwidth, the higher the throughput generated since it s possible to allocate more resource blocks and, thereby, increases the data rate of the user. However, it was found that the spectral efficiency increases only slightly with increasing bandwidth, since the higher the bandwidth, the easier it becomes to manage and respond to user requests. For the CQI value it was found that the higher the value (variable from 1 to 15 depending on the conditions of receipt of the user), the higher the throughput generated. Since the bandwidth used was always the same in these simulations, the spectral efficiency also increases dramatically with the increasing of the CQI value. This is due to the fact that the CQI value is associated with a modulation and a code rate. In are used QPSK, 16QAM and 64QAM modulations and the code rate vary between 0,076 and 0,93. Therefore, the higher the CQI value, the greater the relation modulation/code rate. This makes that the number of bits/symbol increases with increasing CQI value and for this reason, the throughput also increase significantly (from CQI value equal to 1 to CQI value equal to 15 the number of bits/symbol varies between 0,1523 and 5,547 an increase of approximately 37 times). Regarding antennas configurations, uses MIMO technology. This technology allows the use of multiple antennas both emission and reception and through this association is quite possible to increase the data rate. In this technology are specified seven modes (in the case of Release 8), and each one works differently from other modes. With the simulations it was observed that the throughput generated increases with the number antennas (such as spectral efficiency). It was found that the increase is proportional to the function min(ntx;nrx) since if you have a SISO 1x1 configuration and a MISO 4x1 configuration, the throughput generated is about the same. In case you have a SISO 1x1 configuration and a MIMO 4x2 configuration, the throughput generated by the second configuration is approximately twice that of the first one (for MIMO 4x4 configuration would be four times higher). It was also found that in, the maximum number of antennas is 4 (for both emission and reception), however, for -Advanced this number increases to 8 (hence the fact that also achieves better throughput than ). Prospects for Future Work In terms of prospects for future work, it would be interesting to do some simulations, such as simulating the most favorable conditions, both in (Bandwidth = 20 Mhz; CQI = 15; MIMO 4x4), as in the -Advanced (Bandwidth = 20 Mhz; CQI = 15; MIMO 8x8) in order to check whether the results obtained are in agreement with the expected (a spectral efficiency of approximately 15 bit/hz and 30 bit/hz, respectively). It would also be 5 interesting to perform simulations for more than one UE and see what the impact in the throughput generated was. In the case of -Advanced would also be interesting to simulate the carrier aggregation technology and verify if the total throughput generated roughly corresponds to the sum of the throughput generated by each of the carrier combined. References [1] 3GPP. 3GPP - Release 8. [Online] [Last Access: 15 Junho 2011.] http://www.3gpp.org/release-8. [2]. 3GPP - Release 9. [Online] [Last Access: 15 Junho 2011.] http://www.3gpp.org/release-9. [3] ITU - International Telecommunication Union. Radiocommunication Sector (ITU-R) - IMT-Advanced submission and evaluation process. [Online] [Last Access: 11 Junho 2011.] http://www.itu.int/itu-r/index.asp?category=studygroups&rlink=rsg5-imt-advanced&lang=en. [4]. ITU-T IMT-2000 Activities. [Online] [Last Access: 11 Junho 2011.] http://www.itu.int/itu-t/imt-2000/. [5] H. Holma and A. Toskala. for UMTS - OFDMA and SC-FDMA Based Radio Access. 2009. John Wiley & Sons, Ltd.. [6] Agilent Technologies. -Advanced Signal Generation and Measurement Using SystemVue. 2011. Agilent EEsof EDA. Application Notes by Jinbiao Xu. [7] Qualcomm. -Advanced. Mar 2010. Presentation. [8] Nokia. -Advanced research in 3GPP. Dez 2008. GIGA seminar 08 (Presentation by Tommi Koivisto). [9] 3GPP, TS36.213. Evolved Universal Terrestrial Radio Access (E- UTRA); Physical layer procedures. versão 8.5.0. [10] Telesystem Innovations. The Seven Modes of MIMO in. White Papper. [11] A. J. Goldsmith e S. G.Chua. Adaptive Coded Modulation for Fading Channels. Mai 1998. IEEE Trans. on Communications. Vol. 46 pp. 595 602. [12] J. Hayes. Adaptive feedback communications. Fev 1968. IEEE Trans. on Communication Technologies. Vol. 16 pp. 15 22. [13] S. Sesia, I. Toufik and M. Baker. - The UMTS Long Term Evolution - From Theory to Practice. 2009. John Wiley & Sons, Ltd.. [14] Institute of Communications and Radio-Frequency Engineering, Vienna University of Technology. TC: Link Level Simulator. [Online] [Last Access: 27 Maio 2011.] http://www.nt.tuwien.ac.at/about-us/staff/josep-colom-ikuno/ltelink-level-simulator/. [15] MathWorks. MATLAB - The Language of Technical Computing. [Online] [Last Access: 13 Junho 2011.] http://www.mathworks.com/products/matlab/. [16] Agilent Technologies. SystemVue 2011.03. [Online] [Last Access: 05 Junho 2011.] http://www.home.agilent.com/agilent/product.jspx?cc=pt&lc=eng &ckey=1972969&nid=- 34264.973674.00&id=1972969&cmpid=zzfindeesofsystemvue2011_03. [17]. Portugal Home Agilent. [Online] [Last Access: 05 Junho 2011.] http://www.home.agilent.com/. [18]. W1918 -Advanced Baseband Verification Library. [Online] [Last Access: 05 Junho 2011.] http://www.home.agilent.com/agilent/product.jspx?nid=- 34262.967372.00&cc=PT&lc=eng.