Packet scheduling is one of the key features in dataoriented radio interfaces of cellular networks like HSDPA (High Speed Downlink Packet Access). It has been primarily designed for unicast applications. Nevertheless, unicast may not optimise the resource usage when the same content has to be transmitted to several users in the same cell. In this paper, we compare the performance of multicast and unicast scheduling considering both a theoretical generic system and an HSDPA system. We prove the benefit of deploying multicast which is found to have merits when the average channel quality is good enough. Results show that the better the average channel quality is, the more users are allowed to receive the service simultaneously.

Transmission on data-oriented radio interfaces of cellular networks has been primarily designed for unicast applications. Nevertheless, unicast may not optimize the resource usage when the same content has to be transmitted to several users in the same cell. In this context, multicast seems to be an efficient means to convey data. In this paper, we develop an analytical model that allows the computation of the mean bitrate for both multicast and multiple-unicast transmission schemes. Furthermore, we propose a multicast transmission scheme called the equal-bitrate (EB) algorithm that allocates bandwidth to mobiles according to their instantaneous channel quality. We compare it to adaptations of the well-known max-signal-to-noise ratio and round robin to multicast. We propose to group users into clusters. The clustering method combines multicast and unicast transmission schemes according to the user's average channel conditions. We use the analytical model to evaluate the proposed solutions. We compare the resulting performance against pure multicast and multiple-unicast approaches. We show that the EB algorithm offers a good trade-off between throughput and fairness. Also, we show that mixed clustering achieves good performance compared to conventional clustering methods.

This paper targets multicast transmission where the same data is destined to many users simultaneously. Although multicast allows bandwidth saving, it prevents a precise link adaptation over radio links. Indeed, users are subject to the same bitrate despite their different and variable radio conditions. Neither operators nor 3GPP standard offer solutions to support link adaptation in a multicast scenario. In this context, we propose different solutions. We compare conservative and aggressive schemes by computing the resulting throughput performance. For this purpose, we propose a model to compute the average number of retransmissions. We also study the mapping between the reported SNR and the packet sizes in HSDPA systems. We show that the existing mapping offers the best performance for unicast but cuts down the throughput in a multicast scenario. Then, we propose a convenient mapping for multicast, namely shifted mapping. Despite the better precision of this solution, the resulting gain remains marginal.

In a UMTS network, the Radio Access Bearers (RAB) configuration has a direct impact on radio resource usage. The more suited the RAB is to data flows, the more efficient it is. This paper presents an optimal RAB combination for VoIMS (VoIP with IP Multimedia Subsystem in the Core Network). The proposed RAB makes an efficient use of RObust Header Compression (ROHC) which improves the physical layer QoS. The main advantage of the new RAB combination resides in adapting the throughput for SIP (Session Initiation Protocol) signalling and improving the call setup delays for a packet switched transmission. This is achieved via an efficient algorithm of Transport Format selection in conjunction with the flexible rate matching at the physical layer.

No abstract