Comparison of two Access Mechanisms for Multimedia Flow in High Speed Downlink Packet Access Channel Abdelali EL BOUCHTI Abdelkrim HAQIQ Computer, Networks, Mobility and Modeling laboratory e-NGN research group, Africa and Middle East FST, Hassan 1st University, Settat, Morocco Emails: {a.elbouchti, ahaqiq}@gmail.com Abstract— The High Speed Downlink Packet Access (HSDPA) is used more in the radio operator interface of the access network of the Universal Mobile Telecommunication System (UMTS). Its principle is to share a common channel (buffer) with high throughput between the users of the cell in order to improve the Quality of Service (QoS). In this paper, we compare the performance parameters of two access control mechanisms for multimedia flow in HSDPA channel using queue threshold schemes. Real Time (RT) packets (voice and/or video) and Non- Real Time (NRT) packets (data) compose the flow. In the first mechanism, priority in the queue is given to real time packets and a threshold is used to limit the access of these packets in the queue. Another threshold is added in the second mechanism in order to manage NRT packet access in the queue. The RT packets arrive according to a Poisson process and the NRT packets arrive according to a Markov Modulated Poisson Process (MMPP) with two phases. For both mechanisms, different performance parameters (loss probability, average delay of packets, and average number of packets in the queue) have been derived analytically; numerical results are calculated and compared for the two schemes. Keywords- HSDPA, Multimedia Flow, Queueing Theory, QoS, MMPP, Performance Parameters. I. INTRODUCTION While third Generation (3G) wireless systems are being intensively deployed worldwide, new proposals for enhanced data rates and quality of service (QoS) provision are being standardized. One of the most promising enhancements to the widely deployed Universal Mobile Telecommunication System (UMTS) which is based on Wideband Code Division Multiple Access (W-CDMA) is called High Speed Downlink Packet Access (HSDPA) [1] which is also referred to as 3.5G. HSDPA aims at providing data rates around 2-10Mbps on the downlink to mobile users mainly for multimedia services such as real-time and streaming video in packet-switched common channel. In the HSDPA, the link adaptation and packet scheduling functionalities are executed directly from the Node B, this allows advanced packet scheduling techniques. The target of majority of packet scheduling algorithms proposed and studied in the literature [3, 6, 12, 15] is to distribute the time resources between the users. However HSDPA link supports multimedia services, which require differentiated QoS. Thus, besides user priority, the algorithm also has to determine priority among classes supported for a given user. The works in [8, 17] were interested to integrate a buffer per user in node B to take into account intra-user traffic differentiation. Modeling of multimedia traffic over shared is rather complicated. Therefore, most of studies typically investigate the problem using packet-level simulation [2] or as data flows which can be real time (voice or video) or non real time (www browsing, e-mail, ftp, or data access). Other works considered performance study of the HSDPA system taking into account system details rather than the multimedia traffic characteristics [16]. In this paper, we present two access control mechanisms of multimedia flow in HSDPA channel using queue threshold schemes. The RT and NRT packets compose the flow. The thresholds are used in order to manage access packets in the queue giving priority to the RT packets and avoiding the NRT packet loss. The RT packets arrive according to a Poisson process and the NRT packets arrive according to an MMPP process with two phases. In the first mechanism, named without control, both types of packets are not controlled. But in the second mechanism, named with control, the NRT packets are controlled. Our objective is to compare the performance parameters of these mechanisms. The same model of the second mechanism is studied in [11] and enhanced in [10], where the authors modeled the arrival flows by three Poisson processes and they added another threshold. They proved that, the arrival and service rates of the NRT packets don’t have an effect on the performance parameters related to RT packets, whereas those of RT packets influence on all the performance parameters of the system. The authors in [4] and [5] studied a similar model to the second mechanism, they modeled the arrival flow of NRT packets by an MMPP process, because it is capable of capturing inter-flow correlation, and they determined and evaluated the performance parameters. To improve the performance parameters related to the RT packets, recently the authors in [9] proposed an enhanced IJAEST Abdelali EL BOUCHTI, et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES Vol No. 4, Issue No. 2, 029 - 035 ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 29