International Journal of Modern Engineering Sciences, 2013, 2(1): 1-16 International Journal of Modern Engineering Sciences

Transcription

1 International Journal of Modern Engineering Sciences, 2013, 2(1): 1-16 International Journal of Modern Engineering Sciences Journal homepage: ISSN: Florida, USA Article WiMAX Network Performance Improvement through the Optimal Use of Available Bandwidth by Adaptive Selective Voice Coding Bassam F. Gumaidah *, Hasan H. Soliman Dept. of Communication Engineering, Faculty of Engineering, Mansoura University, Egypt * Author to whom correspondence should be addressed; Bassam512360@hotmail.com. Article history: Received 29 November 2012, Received in revised form 5 January 2013, Accepted 7 January 2013, Published 8 January Abstract: In this paper a new method is assumed to improve WiMAX network (carrying voice calls) performance through the optimal use of available bandwidth and this new form is called adaptive selective voice coding (ASVC). This method is compared with several voice codes. The results show that the new method can improves the number of serving customer with saving acceptable voice quality and acceptable packet end to end delay. Keywords: OPNET, Simulation, Voice over IP, WiMAX. 1. Introduction WiMAX is a new technology which offers a lot of internet service such as VoIP, IPTV, and other services. This paper will focus on VoIP service to know how the adaptive selective voice coding(asvc) can improve the usage of available bandwidth of WiMAX network. There are a lot of works try to improve the usage of the available bandwidth of WiMAX by adapting AMC (adaptive modulation and coding) which is a WiMAX qualification inherently or using new algorithm for modulation. For example in reference [2] the authors attempted to use new form of adaptive modulation, and this new form combines with other technique called the simplest Peak to Average Power ratio (PAPR) reduction technique, and they called this new form MC, Modulation adaptation and Clipping. The authors expect this new form has the ability to improve the performance of WiMAX system through reducing the PAPR, improving the SER performance. The idea behind this new form is that assume the data will be transferred through the available bandwidth by different modulation

2 2 scheme at the same time in different percentage depend on the threshold of modulation scheme SNR. For example, at SNR equals to 8 db which is the threshold of BPSK, 45% of the transmitted symbols were mapped with the high order modulation scheme 256-QAM, 15% with 64-QAM, 25% with 16- QAM and 15% of them with 4-QAM. This means 100% of the transmitted symbols were mapped at 8 db with modulation schemes have order higher than the BPSK, and that will apply with other scheme accordingly. Our work will take different approach to improve WiMAX network performance. So this paper will focus on VoIP- through new form called ASVC, adaptive selective voice coding. This new form can optimize the use of available bandwidth, in turn serves more subscribers and accepts more calls with acceptable quality and acceptable packet end to end delay. 2. WiMAX (802.16e) Background WiMAX is a broadband wireless access that supports both fixed and mobile internet access. It is based on IEEE and has maximum data rate of 75Mbits/sec under optimal conditions [3]. WiMAX range covers up to several kilometers. As a result, it can be used for providing wireless broadband across to cities and countries. It can be used as an alternative last mile solution to cable and DSL. WiMAX uses orthogonal frequency-division multiplexing OFDM and scalable orthogonal frequency-division multiple access (SOFDMA). So in this paper, we will perform simulations on these two types of multiplexing to see the effect of each one. WiMAX is based on PHY and MAC layer of the OSI (opens system interconnection) references model (figure 1). The PHY Layer identifies advanced techniques at both modulation and error connection filed [3]. For example, Adaptive modulation and coding (AMC) is a modulation technique allows the base station to change the modulation code from low order modulation code (i.e.qpsk 1 ) to high order modulation code (i.e.64qam 2 ), depend on the distance between the base station and workstations, that allows it automatically increases effective range, when necessary, at the cost of decreasing throughput. As Higher-order modulation provides high throughput at sub maximum range, whereas lower-order modulation provides lower throughput at higher range, from the same base station [3]. Medium access control (MAC) layer of WiMAX uses a scheduling algorithm for the initial entry of the subscriber stations (SS) into the network. Then the base station (BS) allocates an access slot to SS and other subscribers cannot use that slot. The scheduling algorithm is also used for controlling the bandwidth efficiency and quality of service (QoS) parameters by changing the time slot duration based on the SS s application needs. WiMAX uses 2.3 GHz, 2.5 GHz and 3.5 GHz licensed

3 3 bands. Since 2007 WiMAX technology is included in the IMT-2000 set of standards. IMT-2000 standards are defined by the radio communication sector of the International Telecommunication Union (ITU-R). As a result any country that recognizes IMT-2000 standards is able to use WiMAX equipments [4]. Figure 1: OSI Reference model [3]. Hybrid Automatic Repeat Request (HARQ) and Fast Channel Feedback (CQICH) are error correction techniques which are introduced with Mobile WiMAX to enhance coverage and capacity [3]. Table 1 summarizes the coding and modulation schemes supported in the Mobile WiMAX profile the optional UL codes and modulation are shown in italics [5], [6]. Table 1: Supported Code and Modulations [5] DL UL Modulation QPSK,16QAM,64QAM QPSK,16QAM,64QAM Code Rate CC 1/2,2/3,3/4,5/6 1/2,2/3,5/6 CTC 1/2,2/3,3/4,5/6 1/2,2/3,5/6 repetition X2,X4,X6 X2,X4,X6 The combinations of various modulations and code rates provide a fine resolution of data rates as shown in Table 2 which shows the data rates for 5 and 10 MHz channels with PUSC sub-channels. The frame duration is 5 milliseconds. Each frame has 48 OFDM symbols, with 44 OFDM symbols available for data transmission. The highlighted values indicate data rates for optional 64QAM in the UL.

4 4 Table 2 : Mobile WiMAX PHY Data Rates [5]. parameter downlink uplink downlink uplink System bandwidth 5Mhz 10 MHz FFT size Null subcarrier Pilot subcarrier Data subcarrier sub channel Voice over IP Voice is analog and is converted to digital format before transmitting over Internet. This process is called encoding and the converse is called decoding and both are performed by voice Codecs. With bandwidth utilization becoming a huge concern, voice compression techniques are used to reduce bandwidth consumption. Voice compression by a codec adds an additional overhead of algorithmic delay. Thus, a codec is expected to provide good voice quality even after compression, with mini-mum delay. The table 3 shows the bandwidth requirements of some common codecs. Codecs Table 3: Voice codec [11]. Algorithm Bandwidth (Kbps) Ethernet Bandwidth Usage (Kbps) G711 PCM G729 CS-ACELP G723.1 Multi Rate Coder G723.1 Multi Rate Coder G.726 ADPCM G.728 LD-CELP G.711 is the international standard for encoding telephone audio [7]. It has a fixed bit rate of 64kbps. G.723 and G.729 are low bit rate codecs at the expense of high codec complexity. G.723 is one of the most efficient codecs with the highest compression ratio and is used in videoconferencing applications [8]. G.729 is an industry standard with high bandwidth utilization for toll-quality voice calls [9]. G.726 uses ADPCM speech codec standard, and transmits at rates of 16, 24, 32, and 40 kbps. G.728 officially codes speech at 16 Kbit/s using low-delay code excited linear prediction. For example, during a call using G711 as codec, the amount of data transfer for both uplink and downlink will be

5 x 2 = 174.4Kbps = Mbps = Mb per minute. So, G 711 uses Mb/min per VoIP call whereas G 729 uses 0.5MB/min per voice call in the same way [10]. 4. Adaptive Selective Voice Coding (ASVC) The key point of ASVC is that its operation depends on the available bandwidth related to the number of calls or users. When there is enough space of bandwidth (when the number of users or call is small) we can use voice coding with high bit rates (ex: g711), when the usage of bandwidth reaches the maximum space because the number of users or calls increase, the voice coding will change to low bit rates (ex: g729, g723) for the new users. This resulting in the ability to serve more users and save better voice quality than those only use high bit rates voice coding or low bit rates code alone. The proposal logarithm for adaptive selective voice coding: (i) The threshold point with each type of voice code is determined. (Note: the threshold point can be either maximum number of users or maximum throughput; at this paper we use maximum number of users because the maximum throughput is almost the same with all voice code as we work in the same network, but the maximum number of users differs from one voice code to another). (ii) The limit of maximum number can be determined by mean opinion score and packet end to end delay, as when the factor reaches unacceptable value that means it reaches the threshold point. (iii)we use the voice codecs in order from high bite rate g711 to low bit rate g723 in different rate of the maximum number of users. (iv) Step iii can be explained as follow: at first when the free size of bandwidth of the network is large, the users will use high bit rate voice code (g711), when the number of users reach the rate which equal half of the maximum number with g711 voice code, the new users will use first low bit rate g729, when the number of users, which use g729 voice code, reaches quarter of maximum number of users with g729,the new users will use second low bit rate g723. (v) When the network reaches maximum throughputs (use the available bandwidth completely), the network will reject any new user and wait. When any current user exit, its bandwidth will be used by new users, however the new user will use second low bit rate g723, until the network reaches the state as if we use second low bit rate g723 alone. Now the question may be raised, that is, why the second low rate (g723) has not been used from the beginning? The answer is as follows: every voice codec has their advantages and disadvantages, for example, g711 has the highest MOS, in the same time the network cannot serve

6 6 large number of users, on the other side the second low data rate (g723) has toll MOS so the network can serve large number of users. But by using ASVC the network can serve users with high data rate when there is enough space in bandwidth, then use the first and second degrees of low data rate when there are any increasing of the number of users. So by using ASVC the network can serve larger number of user with high MOS value. Figure 2 shows ASVC algorithm. MAX(th)=maximum throughput of the network. N1:#users with G711 at MAX(th) N2:# users with G729 at MAX(th) N3:#users with G723 at MAX(th) N=# users Calculate used bandwidth (current throughput cur(th)) Voice code is G711 NO Cur(th)=>50 % of MAX(th) YES N=>N1 /2 Voice code is G729 YES Cur(th)=>50% & <= 75% of MAX(th) NO N=>(N1/2)+ (N2 /4) Voice code is g723 YES Cur(th)=>75% & <100 %MAX (th) NO N=>(N1/2)+(N2/4)+ (N3/8) Ignore new user until any current user end call Figure 2:ASVC algorithm diagram

7 7 5. OPNET Modeler There are several network simulation program such as Qualnet [12], NS2 (Network Simulation 2) [13], and OPNET modeler. Each of them has its advantages and disadvantages. All of them can be used to simulate wireless network, including WiMAX network. Each program differs from the other, both in terms of the ease of use and the ability to use. OPNET is a research oriented network simulation tool. It is a very powerful software tool that simulates the real world behavior of wired and wireless networks. OPNET Modeler version 14.5 was used in this project for simulating WiMAX links. The OPNET wireless module and the WLAN model provide high-fidelity modeling, simulation, and analysis of wireless networks, including the RF environment, interference, transmitter / receiver characteristics, and full protocol stack, including MAC, routing, higher layer protocols and applications. Furthermore, the ability to incorporate node mobility and interconnection with wire-line transport networks provide a rich and realistic modeling environment [14]. OPNET MODELER was selected, as it has the following abilities: a) Provides a comprehensive development environment supporting the modeling of communication networks and distributed systems. b) Performs discrete event simulations. c) Provides Graphical specification of model wherever possible; so models are entered via graphical editors. d) Has library of models for most of the common networks around us. The OPNET WiMAX Specialized Model is available for OPNET Modeler Wireless Suite and OPNET Modeler Wireless Suite for Defense. It supports the IEEE and IEEE e standards [15]. 6. Simulation OPNET modeler 14.5 is used to simulate the WiMAX network as shown in figure 3. The concerned performance metrics are as followings: (i) Jitter: describes the degree of variability in packet arrivals. Jitter delay tolerate can be 75 milliseconds (40 milliseconds is preferred) [16]. (ii) Packet end to end delay: is the average time it takes for a packet to travel from its source to its destination. According to ITU recommendation of G.114, the maximum amount of packet end to end delay that a voice call can tolerate one way is 150 milliseconds (100 milliseconds is preferred).

8 8 (iii)throughput: is a measure for the pure data which we can transfer successfully through the network. (iv) Signal to noise ratio (SNR): is a measure of signal strength relative to background noise. (v) Path loss: is the reduction in power density (attenuation) of an electromagnetic wave as it propagates through space. (vi) Mean Opinion Score (MOS): it provides a numerical indication of the perceived quality of received media after compression and/or transmission. It is expressed as a single number in a scale of 1 to 5, where 1 is the lowest perceived quality, and 5 is the highest perceived quality [17]. Figure 3: Network The simulation is implemented through two steps as described below: Step1: (i) Multiple Scenarios for each voice code (g711, g729 and g723) are implemented. The first scenario is implemented with 8 workstation make call with each other through base station. (ii) Then the scenario is repeated with increase on the number of workstation until reach the threshold point which is mentioned before. (iii)the threshold point can be determined through the study of several factors such as Mean Opinion Score (MOS), packet end to end delay, WiMAX throughput, work station throughput, work station packet end to end delay. (iv) The study will conclude the maximum number of users which can be served with each voice code.

9 9 Step 2: (i) The results form state 1 which will be used to implement ASVC algorithm. Figure 3 is a simple form for the studied WiMAX network. The simulation is implemented with duration of300 seconds, and the following results will be shown: (i) Show the MOS, Packet End to End Delay, base station throughput, work station throughput, and work station packet end to end delay with G711, G723 and G729 voice coding. (ii) Show the MOS, Packet End to End Delay, base station throughput, work station throughput, and work station packet end to end delay with ASVC (adaptive selective voice coding). Table 4 lists the network setting for all nodes: a. Base station b. Workstation c. WiMAX configuration Table 4: Network configuration parameter Base station Work station Antenna gain db 15 db -1 db Transmitted power 38 W 0.5 W PHY profile Wireless OFDMA 20 MHz Wireless OFDMA 20 MHz PHY profile type OFDM OFDM Efficiency mode Physical layer enable Frame duration 5 msec Symbol duration 102 msec Number of subcarrier 2048 Duplexing technique TDD Base frequency 5.8 GHz Channel bandwidth 20 MHz WiMAX configuration 7. Results and Discussion OPNET modeler can give us the result as graphs. This graphs form a relation between the studied factor and simulation times or other effected factor like number of users. Also these results can be converted to excel data sheet to deal with them. The simulation duration is 300 seconds and the scenario is repeated with increase the number of workstations from 8 to 412 for first state, for second state the number of workstation is increased from 8 to 230.

10 10 As mentioned before the simulation is implemented through two states, so the results will be divided to two states: Step 1 will show us the following factors: MOS, Packet End to End Delay (PETED), base station throughput, work station throughput, and work station packet end to end delay for each voice code as the following figures: Figure 4: Step 1 MOS Figure 5: Step 1 PETED

11 11 Figure 6: Step 1 global throughput Figure 7: Step 1 workstation throughput

12 12 Figure 8: Step 1 workstation PETED 7.1. Discussion The threshold point can be determined from figure 4 and figure 5. It is an evident at this point the performance of network is not acceptable, as the MOS go down to poor value and the packet end to end delay go up to high value which is not acceptable for voice communication. Figure 6 shows the global throughput. It is noticed that after the threshold point the global throughput stays fixed, and that explains why the work station throughput goes down after threshold point (as we can see it in figure 7), as when the network reach the threshold throughput any new user will leads to data drop, that drop will occurs on all work stations on the network so the workstation throughput goes down with each new user been added to the network. Table 5 shows the simulation results. Table 5 : Summarized result Voice code G711 G729 G723 Maximum number of users n MOS (n) PETED (n) (second) G. Throughput(n) (Bits/second) WS. Throughput (Bits/second) WS. PETED sec

13 13 Step 2 will show us the following factors: MOS, Packet End to End Delay, base station throughput, work station throughput, and work station packet end to end delay for ASVC (adaptive selective voice coding) in the following figures: Figure 9: ASVC MOS Figure 10: ASVC PETED

14 14 Figure 2: ASVC Throughput It is noticed from figure 9 that ASVC has a smooth transition in MOS with increasing the number of users, not sharp transition which is noticed with the codes in first state. Also ASVC presents slow increase in packet end to end delay which is noticed from figure 10. The following figures show MOS and PETED for workstation during 300 seconds of simulation. Figure 3 : ASVC workstation MOS

15 15 Figure 4: ASVC workstation PETED Table 6 shows the results of ASVC with 200 workstations. Table 6: ASVC summarized results ASVC number of users n 200 Global MOS (n) ~3.65 Global PETED (n) second 0.6 G. Throughput(n) Bits/second WS. MOS WS. PETED second Conclusion It is concluded that when each voice codes is used alone a good performance appears on one hand, but a bad effect on other. For example with high data rate G711 voice code 64 Kbit/s we can get good quality of voice but the maximum number of users cannot exceed 60. Beyond 60 users, a large value of packet end to end delay occurs. On the other hand with low data rate voice code G Kbits/sec, the number of users can reach up to 350 users, but with low quality of voice comparing with G711. With ASVC method, more users can be served with acceptable quality of voice and acceptable packet end to end delay when it is compared with other voice codes.

16 16 References [1] Jeffrey G. Andrews, Arunabha Ghosh, and Rias Muhamed, Fundamentals of WiMAX: Understanding Broadband Wireless Networkin, Prentice Hall, [2] Ibrahim Ismail Al-kebsi, Mobile WiMAX Performance Improvement Using a Novel Algorithm with a New Form of Adaptive Modulation, IJCSNS International Journal of Computer Science and Network Security, February(2009): [3] Deepak Pareek, WiMAX: Taking Wireless to the MAX, in WiMAX: Taking Wireless to the MAX.: Taylor & Francis Group, LLC, p , [4] Will Hrudey, Streaming Video Content Over IEEE / WiMAX Broadband Access, [5] Mobile WiMAX Part I: A Technical Overview and Performance Evaluation, Aug [6] Mohammad Tawhidul Alam Mohammad Saiful Islam, WiMAX: An Analysis of the existing technology, [7] ITU-T. Recommendation G.711. [Online]. rec/t-rec-g.711/e [8] ITU-T. Recommendation G.723. [Online]. rec/t-rec-g.723/e [9] ITU-T. Recommendation G.729. [Online]. rec/t-rec-g.729/e [10] Whirlpoo. [Online]. [11] Cisco. Voice Over IP - Per Call Bandwidth Consumption. [Online]. [12] QualNet. [Online]. [13] The Network Simulator - ns-2. [Online]. accessed] May, [14] opnet. [Online]. [15] OPNET. [Online]. [16] F.P. Zhang, O.W.W. Yang, and B. Cheng, "Performance evaluation of jitter management algorithms," in Proc. of Canadian Conference on Electrical and Computer Engineering, 2(2001): [17] JERRY D.GIBSON and EDITOR, Multimedia Communications. Southern Methodist University, Dallas,Texas, 2001.