On the Optimizing of LTE System Performance for SISO and MIMO Modes

Transcription

1 2015 Third International Conference on Artificial Intelligence, Modelling and Simulation On the Optimizing of LTE System Performance for SISO and MIMO Modes Ali Abdulqader Bin Salem, Yung-Wey Chong, Sabri M. Hanshi National Advanced IPv6 Centre Universiti Sains Malaysia USM, Penang, Malaysia {ali, chong, Tat-Chee Wan National Advanced IPv6 Centre and School of Computer Sciences Universiti Sains Malaysia USM, Penang, Malaysia Abstract size and modulation scheme can play an important role on optimizing the data rate of Long Term Evolution (LTE) system. The performance of the existing LTE system is evaluated in this paper in term of throughput and packet loss for Orthogonal Frequency-Division Multiple Access (OFDMA) modulation scheme with different packet sizes. The performance of LTE system for different packet size is done. The simulation experiments are done using Matlab and Simulink libraries. The results show and discuss the effects of packet size and adaptation modulation on the performance of LTE based on throughput and packet loss. Moreover, the Signal-to-noise ratio () threshold values can be optimized in order to enhance the performance of LTE system. Experiments of packet loss illustrated that it is possible to get an accepted Error Rate (PER) before reaching to threshold values by enhancing an optimizing technique such as Markov Decision Process (MDP). Keywords- LTE system; error rate; Signal-to-noice ratio; Throughput and packet loss; SISO; MIMO I. INTRODUCTION LTE is the latest standard evolution of Global System for Mobile communication / Universal Mobile Telecommunications System (GSM/UMTS). LTE is a project of the 3rd Generation Partnership Project (3GPP) group. According to Global Mobile Suppliers Association (GSA), a 274 LTE networks had been launched in 101 countries until 2014 Feb [1]. The goal of LTE is to increase the speed and capacity of wireless networks by using signal processing techniques and modulations [2]. Long Term Evolution-Advanced (LTE-A) is the most recent project that addresses 4G technology requirements. LTE and LTE-A become an excellent solution for the deployment in places where it is difficult to get with other technologies such as cable or Digital Subscriber Line (DSL) because of the costs of deployment, maintenance of such technologies, as well as the added specific features that LTE is distinguished by, such as the ability of reaching to 20MHz and 100 MHz transmission bandwidth of LTE and LTE-A respectively [3]. Such bandwidth can increase the peak data rate to 100 Mbps and 1000 Mbps for Down Link (DL) LTE and DL LTE-A respectively [4]. Presently, the data carried by wireless network devices are merged with the multimedia system [5]. This has gravitated the cellular industry on providing higher data rates which led to consider the video applications as one of the key applications in this network mode [6]. Achieving higher data rate can simply be done by optimizing the packet size, so the video data can be transmitted in rates as high as possible with less error rate. The typical packet in LTE network can reach to 1500 byte including payload and headers of LTE protocol layers. However, the 3GPP standards have not speci ed any LTE fixed packet size that can be used as optimum size to deliver the video data. In addition, many researchers contributed and enhanced methods to get optimum higher data rate. Hence the need to evaluate the performance of existing LTE system is necessary. In this paper, the performance of LTE is evaluated in terms of throughput and packet loss for OFDMA modulation scheme and different packet sizes. The aim of this work is to investigate the adaptation modulation scheme of the existing LTE system and its applicability to enhance the performance of LTE networks for different packet sizes which will be presented below. Section II discussed how the PER and threshold values are determined mathematically. Beside that the threshold values were validated by determining an estimation average threshold values. These values were used to get the experiments of throughput and packet loss for different modulation types with different packet sizes. Furthermore, system design and simulation environments of LTE system were presented in sections III and IV respectively. Experiments of LTE systems such as throughput and packet loss were discussed in section V. Finally, the Discussion section of the experiments as well as the Conclusion were discussed in sections VI and VII respectively. II. SETTING PER AND THRESHOLD VALUES FOR VARIOUS PACKET SIZES A. PER The relationship between the BER and PER is valid for an ideal communication system that transmits data over binary symmetric channel with uncorrelated noise. The PER /15 $ IEEE DOI /AIMS

2 is determined exclusively by the BER and the number of bits in the packet s data payload. The length of the packet in bits is and the bit error probability for the channel is. Any packet that exceeds the BER threshold value will be discarded. The probability PER of getting a packet in error is given by [7]: (1) As a packet, the PER must not exceed 10%, hence: (2) Therefore, the probability of receiving x correct packet should exceed 90%, hence: (3) can be expressed as: (4) To determine how much BER corresponding to 10% of PER is given by: (5) (6) For instance let L=8000 then (7) So in order to receive packet correctly: (8) (9) TABLE I. REPRESENT VARIOUS ACCEPTED BER VALUES OF PER FOR DIFFERENT PACKET SIZES size in bits Maximum accepted BER of 10% PER B. Treshold Values threshold values for AWGN could be determined as: (10) (11) (12) For conversion from watt to db, it will be: (13) (14) threshold values are determined mathematically based on modulation type and BER. To validate the threshold values of that are gotten from equations (14), a simulation is run with different initial seeds to generate various noises. At the end, the estimated average threshold values are determined based on modulation type and BERs that are related to different packet sizes (see Table 1) as shown in Table 2. The purpose of this experiment is to: Measure and validate how the mathematical and simulation experiments are close to each other. Use the values that were obtained from the simulation experiment as threshold values to get the experiments of throughput and packet loss for different modulation types with different packet sizes. Table 2 represents that the threshold values of mathematical and practical experiments are close to each other. These values are going to be a little bit high in the practical experiment when having a packet size of 8000 bits. This is due to the fact that when the packet size is increased, the total packet numbers in the frame is decreased and hence the probability of achieving a 10% in PER will be higher. Consequently, the required becomes a little bit high to ensure that the packets are in the range of accepted PER. TABLE II. THEORETICAL AND PRACTICAL VALUES FOR DIFFERENT PACKET SIZES III. SYSTEM DESIGN LTE simulation involves a data channel from a basestation (enode B) to a mobile user equipment (UE). It implements transmission mode 4 (closed loop) of the LTE standard, which is specified in 3GPP TS v10.7.0, Closed loop system requires channel knowledge at the transmitter, and it occurs when access network executes dynamic adjustment based on the UE s feedback. The connection between enodeb and UE is PTP with a MIMO Channel Model. The summarized parameters of the LTE simulation are illustrated in Table

3 IV. SIMULATION ENVIRONMENTS The simulation experiments were done using Matlab and Simulink libraries, which are discrete events simulation [8], thus they are suitable to simulate the proposed model. The simulation was performed in Lenovo with Intel Core i7-4700mq 2.40 GHz, 12 GB RAM. TABLE III. PARAMETERS SETTING OF LTE SIMULATION Parameter Carrier Frequency Sub-carrier Spacing Transmission Bandwidth Transmission Mode Channel Model MIMO Channel Antenna Schemes MAC Header Type and Setting GHz 15 khz 20 MHz Closed loop (TM4) AWGN / Rayleigh Static MIMO, ETU MIMO 2 2 QPSK, 16-QAM, 64-QAM R/R/E/LCID/F/L (2 Byte) V. EXPERIMENTS OF LTE SYSTEM This section discusses the experiments of existing LTE system based on throughput and packet loss. Consequently, the simulation was run 20 times with 6000 frames for each run. The average experiments throughput and packet loss of the system are analyzed based on SISO and MIMO modes and on different packet sizes (see Table 1) by taking into consideration that the MAC header is included in the specified packet sizes, so when throughput is calculated, the MAC header is excluded from the data. A. Throughput Measurement In this section, the throughput experiments were done and categorized into two main scenarios based on SISO and MIMO modes. In addition, the experiments were done for different packet sizes. At the end, the corresponding data rate of threshold values for different packet sizes were highlighted. In SISO mode, the throughput measurements are obtained in Fig 1(a,b,c,d) for various packet sizes (see Table 1), with the corresponding threshold values that are required to deliver the data. For example, in case of 1000 bits as packet size, Fig 1(a) shows that the threshold values (12.2, 18.85, 25 db) are corresponding to the throughput values (22.9, 45.5, 69 Mbps) for QPSK, 16-QAM, and 64-QAM respectively. Table 4 summarizes the experiments in Fig 1(a, b, c, d), where the threshold values with corresponding throughput are represented for various packet sizes. Similar experiments were done, for 2x2 MIMO mode with different packet sizes which illustrated the number of received bit for the required threshold values as shown in Fig 2 (a, b, c, d) and Table 5. Compared to the corresponding throughput in SISO mode, the throughput in 2x2 MIMO mode is almost double for most packet sizes [3]. As shown in figures 1 and 2, the corresponding throughput of the threshold values are increased when the packet size is increased unless it reaches 8000 bits per packet while the modulation is in 64-QAM, where its throughput values are Mbps and 139 Mbps in SISO and MIMO modes. This exceptional case is slightly decreased compared to 2000 and 4000 bits as packet size because the packet size of 8000 bits is very sensitive to packet loss which affects directly the throughput. TABLE IV. AND CORRESPONDING THROUGHPUT VALUES FOR VARIOUS PACKET SIZES OF SISO MODE IN THE LTE SYSTEM TABLE V. AND CORRESPONDING THROUGHPUT VALUES FOR VARIOUS PACKET SIZES OF MIMO MODE IN THE LTE SYSTEM B. Loss Measurement This section presented the scenario of the accepted packet loss for a given threshold values in different packet sizes. The scenario has three different experiments based on modulation type, and for each modulation, experiments are presented for different packet sizes (see Table 1). Fig 3 illustrates the packet loss at different threshold values. The packet loss is decreased when the packet size is increased. But it differs when the packet size is at 8000 bits while the modulation is 64-QAM, this is because its size is very sensitive to the packet loss and this will cause the packet loss to be dropped down rapidly from unacceptable level at (> 10%) to acceptable level at (<< 10%). VI. DISCUSSION As shown in figures 1, 2, 3 where the specified threshold values stand, the link adaptation can be switched from one modulation type to another by taking into consideration the accepted PER when the data is transmitted and hence, throughput is increased. Also in general, when packet size is increased, the throughput at specified threshold values are increased too. In addition, as shown in fig 3(b) for example when the packet size stand at the point of 1000 bits, the PER is which is corresponding to db. Similar findings can be reached for the other packet sizes unless packet size reaches 8000 bits while the modulation is at 64-QAM. This will result the 10% PER, which is gained at db. In general, it is illustrated that it is possible to get an accepted PER before reaching threshold values. This finding will lead to contribute other 414

4 models for achieving link adaptation values by enhancing an optimizing technique such as MDP, and hence increasing the throughput with less value. Figure 2. (a, b, c, d) Throughput vs for Various s and Modulation of MIMO Mode in the LTE System Figure 1. (a, b, c, d) Throughput vs for Various s and Modulation of SISO Mode in the LTE System 415

5 VII. CONCLUSION In conclusion, this study provides a brief overview of LTE networks and discusses the needs to evaluate the performance of LTE system. In addition, determining the PER and threshold values were presented mathematically and practically. Furthermore, system design and simulation environments are presented too. Finally, experiments of LTE system and their discussion are presented in term of throughput and packet loss. Experiments of packet loss illustrated that it is possible to get an accepted PER before reaching the threshold values by enhancing an optimizing technique such as MDP. ACKNOWLEDGMENT The research project is funded by Universiti Sains Malaysia under grant No.: 304/PNAV/ Figure 3. (a, b, c) Loss of Threshold Values for Different s and Various s in the LTE System REFERENCES [1] Y. C. Sung and Y.-H. Yang, "Challenges from voice-over-lte to video-over-lte," in Network Operations and Management Symposium (APNOMS), th Asia-Pacific, 2014, pp [2] I. Motorola, "Long Term Evolution (LTE): A Technical Overview," [3] A. A. Bin-Salem, T.-C. Wan, Y.-W. Chong, and I. J. Mohamad, "LTE Peak Data Rate Estimation Using Modified alpha-shannon Capacity Formula," in Proceedings of the AINTEC 2014 on Asian Internet Engineering Conference, 2014, pp [4] 3rd Generation Partnership Project 3GPP TS v11.1.0, "LTE; Evolved Universal Terrestrial Radio Access (E-UTRA); Multiplexing and channel coding (Release 11)," [5] A. Bin-Salem and T. Wan, "Survey of Cross-layer Designs for Video Transmission Over Wireless Networks," IETE Technical Review, vol. 29, pp , [6] N. Saxena, S. Singh, A. Roy, and D. H. Ail, "NEST: novel embms scheduling technique," Wireless Networks, pp [7] D. G. Yoon, S. Y. Shin, W. H. Kwon, and H. S. Park, " error rate analysis of IEEE b under IEEE interference," in Vehicular Technology Conference, VTC 2006-Spring. IEEE 63rd, 2006, pp [8] D. A. Tacconi and F. L. Lewis, "A new matrix model for discrete event systems: application to simulation," Control Systems, IEEE, vol. 17, pp ,