Adaptive Modulation and Coding Simulations for Mobile Communication Networks Nizar Zarka, Amoon Khalil and Abdelnasser Assimi Higher Institute for Applied Sciences and Technology Communication Department Damascus, Syria Email: nizar.zarka@hiast.edu.sy Abstract—This paper presents the simulations of Adaptive used are QP SK, 16 − QAM or 64 − QAM . The modulated Modulation and Coding (AMC) for Mobile Communication Net- complex symbols are sent to an Additive White Gaussian works. The simulations show that AMC gives higher throughput Noise channel (AWGN). At the receiver a soft demodulation is than the static modulation and coding with a gain of 4dB of the Signal-to-Noise Ratio (SNR). The simulation results are applied to get LLR followed by a de-puncturing where zeros validated using real measurements of the High Speed Packet are added to the removed symbols during the puncturing phase. Access Evolution (HSPA+) mobile network. Finally symbols are detected in the Turbo decoder Vn . In the following we will describe each part of the system. I. I NTRODUCTION The increasing demand of mobile multimedia services in- cluding VoIP, mobile TV, audio and video streaming, video conferencing, FTP and internet access require intelligent com- munication systems able to adapt the transmission parameters based on the link quality. Changing the modulation and coding scheme yield a higher throughput by transmitting with high information rates under favorable channel conditions and reducing the information rate in response to degradation effects of the channel [1]. The idea behind the Adaptive Modulation and Coding (AMC) is to dynamically change the modulation and coding scheme to the channel conditions. If good Signal-to-Noise Ratio (SNR) is achieved, system can switch to the highest order modulation with highest code rates (e.g. 64−QAM with code rate CR = 34 ). If channel condition changes, system can shift to other low order modulation with low code rates (e.g. QP SK with CR = 12 ) [3]. The aim of this paper is to understand how AMC works, to develop our own simulation tools and validate the simulation results with the real measurements of the HSPA+ mobile network [2]. The paper is organized as follows; first we present the design of the proposed system model for mobile communication networks, followed by the simulation results of the static and the adaptive modulation and coding and the validation from real measurements and finally a conclusion. II. P ROPOSED S YSTEM M ODEL Figure 1 represents the simulation model used in our simula- Fig. 1. AMC Proposed System Model tion program. It is composed of a transmitter, a communication channel and a receiver. At the transmitter Un data are encoded A. Turbo Encoder to get Cn using Turbo code with rate 31 . Some symbols are The Turbo encoder [5] is composed of two identical Recur- removed in the puncturing block to give Dn depending on the sive Systematic Convolution (RSC) as it shows in Figure 2. code rate CR = 13 , CR = 12 or CR = 23 [4]. The modulation The two coders receive the same input data Uk with reordering via the interleaver. The outputs are composed of three symbols Copyright c 2016 held by the authors. Uk , Pk1 and Pk2 . The code rate of the turbo encoder is CR = 31 . 36 QPSK, 16−QAM and 64−QAM modulations. QP SK with CR = 21 gives 2 bits per symbol, 16-QAM with CR= 34 gives 4 bits per symbol and 64 − QAM with CR = 34 gives 6 bits per symbol. Fig. 2. The Turbo Encoder Fig. 6. Distribution of symbols in QP SK B. Puncturing Puncturing [6] is applied at the outputs of the turbo encoder, to groups of turbo code symbols to reduce the coding rate of the transmitter. The puncturing Matrix contains 1 and 0 to give different code rate CR = 13 , CR = 12 and CR = 32 as it shows in Figure 3, Figure 4 and Figure 5. CR = 13   1 1 1 1 1 1 1 1 P = 1 1 1 1 1 1 1 1  1 1 1 1 1 1 1 1 Fig. 3. Puncturing matrix with CR = 13 Fig. 7. Distribution of symbols in 16 − QAM CR = 12   1 1 1 1 1 1 1 1 P = 1 0 1 0 1 0 1 0  0 1 0 1 0 1 0 1 Fig. 4. Puncturing matrix with CR = 12 CR = 23   1 1 0 1 1 0 1 1 P = 0 0 0 0 1 1 1 1  0 1 1 1 0 1 0 1 Fig. 5. Puncturing matrix with CR = 23 C. Modulation Fig. 8. Distribution of symbols in 64 − QAM In the M − QAM modulation symbols in the inputs are divided into blocks of length k = log2 (M ), where M is D. The Receiver the rank of the modulation. Each block is connected to one The channel used is AWGN [8]. At the receiver symbols point of the M − QAM modulation [7]. Figure 6, Figure 7 are demodulated and sent to the decoder which is composed and Figure 8 show respectively the distribution of symbols in of de-puncturing and Turbo decoder. The de-puncturing uses 37 the same puncturing matrix to redistribute the symbols in the correct place before the puncturing. The deleted symbols are replaced with zeros. The de-puncturing gives the values to the turbo decoder which concludes the sent symbols [9]. In the following we will explain the results of the simulations. III. S IMULATION R ESULTS OF THE S TATIC M ODULATION AND C ODING The simulations have been run in MATLAB under AWGN and Rayleigh fading channels [10]. We first present the perfor- mance of different types of modulations and coding, then we conclude the table of the threshold of the SNR for the each modulation and coding. Finally we present the throughput in Fig. 10. 16 − QAM Performance term of SNR in AWGN channel and Rayleigh fading. A. The Performance of QPSK C. The Performance of 64-QAM Figure 9 depicts the Bite Error Rate (BER) in term of SNR with the scenarios of QPSK modulation without and with code Figure 11 depicts the BER in term of SNR with the rates of CR = 13 , CR = 12 , CR = 23 . It is shown that the scenarios of 64 − QAM modulation without and with code best performance occurs for CR = 31 , and the performance rates of CR = 13 , CR = 12 , CR = 23 . It is shown that the decreases with the increase of CR. The figure shows that best performance occurs for CR = 31 , and the performance for a fixed value of BER = 10−4 , the gain in SNR with decreases with the increase of CR. The figure shows that coding, comparing to the modulation without coding, is equal for a fixed value of BER = 10−4 , the gain in SNR with to 5dB, 8dB and 12dB when CR = 23 , CR = 21 , CR = 13 coding, comparing to the modulation without coding, is equal respectively. to 5dB, 10dB and 15dB when CR = 23 , CR = 12 , CR = 31 respectively. Fig. 9. QP SK Performance Fig. 11. 64 − QAM Performance B. The Performance of 16-QAM Figure 10 depicts the BER in term of SNR with the D. Comparison of Performances scenarios of 16 − QAM modulation without and with code rates of CR = 31 , CR = 12 , CR = 23 . It is shown that the best performance occurs for CR = 31 , and the performance The figure 12 shows that the performance of 16 − QAM decreases with the increase of CR. The figure shows that with CR = 31 is better than QP SK with CR = 32 , though the for a fixed value of BER = 10−4 , the gain in SNR with number of bits per symbol is 13 in both case. The performance coding, comparing to the modulation without coding, is equal of 64 − QAM with CR = 13 is better than 16 − QAM with to 5dB, 9dB and 13dB when CR = 23 , CR = 21 , CR = 13 CR = 13 , though the number of bits per symbol is 2 in both respectively. cases. 38 Fig. 12. Performances of different modulation with different code rates Fig. 15. Throughput in term of SNR in AWGN channel Figure 13 shows the Frame Error Rate (FER) in term of SNR for different modulations and coding. Similar results are obtained in Figure 16 that shows the throughput in the presence of AWGN channel and Rayleigh fading. Fig. 13. Frame Error Rate in term of SNR Figure 14 shows the minimum thresholds of SNR that allows to choose the respective modulation and coding for F ER = 10−2 . We notice that 16 − QAM with CR = 13 Fig. 16. Throughput in term of SNR in AWGN channel with Rayleigh fading replaces QPSK with CR = 23 , and 64 − QAM with CR = 13 replaces 16 − QAM with CR = 12 . IV. S IMULATION OF THE A DAPTIVE M ODULATION AND bpsym SN R(dB) CR M odulation C ODING 2/3 − 1/3 QP SK 1 3 1/2 QP SK 4/3 4 1/3 16 − QAM Our simulation of the adaptive modulation and coding is 2 9 1/3 64 − QAM shown in Figure 17. The block diagram is similar to the Figure 8/3 12.5 2/3 16 − QAM 1 with the additional Adaptation block. The demodulator at 3 14.5 1/2 64 − QAM the receiver part calculates the SNR and sends it to the 4 18.5 2/3 64 − QAM Adaptation block which decides the suitable modulation and Fig. 14. The Minimum threshold for modulation and coding coding. Figure 18 shows the curves of the throughput in term of SNR for the adaptive modulation and coding with Rayleigh fading. The curve matches the curve of QP SK and CR = 13 E. Throughput in term of SNR for AWGN and Rayleigh fading for low values of SNR, and matches the curves of 64 − QAM Figure 15 shows the throughput in the presence of AWGN and CR = 23 for high values of SNR. For the medium value channel. We notice that 16−QAM with CR = 13 gives higher of SNR we notice a gain of 4dB at SN R = 15dB. By throughput than QP SK with CR = 23 . We also notice that consequence the throughput curve of the adaptive modulation 64 − QAM with CR = 31 gives higher throughput than 16 − and coding acts like an envelope of the maximum values of QAM with CR = 31 . the throughput of the static modulation and coding. 39 Fig. 19. Throughput of AMC Figure 19 shows the measured and the simulated throughput in term of SNR for the adaptive modulation and coding. We noticed that most of the points of the measurements are located near the throughput of the adaptive modulation and coding. Some points deviate because of the difference between the real channel and the simulated channel. VI. C ONCLUSION This paper presented the simulations of the adaptive mod- ulation and coding for mobile communication networks. We Fig. 17. Throughput of AMC with Rayleigh channel concluded that the throughput could be increased by increasing the modulation order and the coding rate with the increase of Signal-to-Noise Ratio. The AMC gives higher throughput by changing the modulation and coding in function of the Signal-to-Noise ration at the receiver with a gain of 4dB. The simulation results are validated via drive test measurements of HSPA+ mobile network. R EFERENCES [1] S. Salih and M. Suliman, Implementation of Adaptive Modulation and Coding Technique using, International Journal of Scientific and Engi- neering Research Volume 2, Issue 5, May-2011, ISSN 2229-5518. [2] S. Kottkamp, HSPA+ Technology Introduction, Application Note, Rohde and Schwarz, 2009 [3] S. Hadi and T. Tiong, Adaptive Modulation and Coding for LTE Wireless Communication, IOP Conf. Series: Materials Science and Engineering 78, 2015. [4] M. Fernandez and M. Rupp and M. Wrulich HSDPA CQI Mapping Optimization Based on Real Network Layouts, Bachelor Thesis, Institute of Communications and Radio-Frequency Engineering, 2008. [5] F. Huang T. Pratt, Evaluation of Soft Output Decoding for Turbo Codes, Master of Science, Electrical Engineering,ETD etd-71897-15815, 1997. [6] M. Tuechler and J. Hagenauer. Channel coding, lecture script, Munich Fig. 18. Throughput of AMC University of Technology, pp. 111-162, 2003. [7] J. Proakis, Digital Communications, fourth edition, Mc-Graw Hill, 2001 [8] P. Farrell and J. Moreira, Essentials of Error-Control Coding, John Wiley V. VALIDATION OF THE AMC S IMULATION RESULTS and Sons Ltd, 2006. [9] K. Bogawar, S. Mungale and M. Chavan, Implementation of Turbo To validate our simulation results of the AMC, we use Nemo Encoder and Decoder, International Journal of Engineering Trends and Technology (IJETT), Volume 8 Number 2- Feb 2014. outdoor and Nemo Analyze tools [11] from one of the mobile [10] B. Sklar Rayleigh Fading Channels in Mobile Digital Communication operator in the country. The tools allow to measure, store and Systems, Part I: Characterization,IEEE Communications Magazine, July analyze the parameters of the HSPA+ network between the 1997. [11] T. Nugraha Drive Test Nemo,Technology, Business, Aug 19, 2012. Node B and the user (e.g. modulation and coding, Energy chip/Noise, Channel Quality Indicator, Number Channeliza- tion Code). 40