Performance Modeling of the Fast and Reliable Two-Layer Retransmission for CDMA Systems

资源类型:pdf 资源大小:839.00KB 文档分类:工业技术 上传者:王月玥

相关文档

批量下载下列文档

文档信息

【出版日期】2005-03-25

【刊名】The Journal of China Universities of Posts and Telecommunications

全文阅读

1 Introduction With the proliferation of the World Wide Web (WWW)in our daily life, a number of wireless data services[1] suchas voice, audio, video streaming, file and web downloadingalso need to be supported in the wireless access networks.To bring the WWW traffic to the wireless mobile devices, itis important that a suitable protocol or standard is chosen tocater to the growing demands of data services over wirelesschannels which could handle a wide variety of multimediatraffic with different Quality of Services (QoS) require ments. No matter which wireless data technology emerges asthe dominant radio interface, the fact remains that it will re ly on the IP based network because Internet Protocol(IP) iscurrently the most dominant inter networking protocol.Transport Control Protocol (TCP), the commonly used trans port layer protocol for IP, is still the major suite for IP andprovides reliable end to end transmission[2] in the wirelinedomain. The design of TCP has been such that it performswell in wireline networks where the channel error rates areextremely low and any occurrence of congestion is due to lossof packets only. However, when TCP is used in the wirelessdomain, which is characterized by high bit rate, the perfor mance of TCP severely degrades[7]. Any packet loss at thewireless link is interpreted as congestion by TCP, which re sponds to it by reducing the transmission window size, initi ating the congestion control mechanism and resetting the re transmission time[3]. The congestion control mechanism de signed for wireline networks causes an unnecessary reductionin the TCP throughput. Several schemes have been proposedto alleviate the effects of non congestion related losses overwireless links[4~6]. Radio Link Protocol (RLP) is one such scheme thatshields the effect of the loss over wireless links from the TCPlayer[7~8]. The RLP is generally employed between thephysical layer and the TCP layer. It fragments an upper lay er packet (a TCP segment in this case) into several RLPframes before transmitting over wireless channel. In case of aRLP frame loss during transmission, the RLP uses an Auto matic Repeat reQuest (ARQ) error recovery mechanism toretrieve the lost RLP frame. The process for recovery of erro neous frames is initiated by the receiver by requesting re transmission of the missing or damaged frames. Radio linkprotocols are usually sufficient to shield the physical layerimpairment from the TCP, but might fail to do so if the ap plication has very strict delay requirements. The delay asso ciated with retransmission at the RLP might not be small e nough to sustain a real time application. To tackle this prob lem, another fast retransmission mechanism must be incorpo rated below the RLP layer. This can be done by an ARQmechanism at the MAC layer, thus providing two layers ofretransmission reliability. The benefit of MAC layer ARQ isthat retransmissions can be done very quickly without notify ing the upper RLP layer. Meanwhile, other technology linksoft combing can be used to improve the probability of cor rectly receiving the frames in MAC layer. Some studies[11~15] mainly consider the effect of RLP lay er on TCP performance. Other researches[9~10] about the twolayers retransmission focus on their joint reliability offered bythe MAC and RLP for CDMA systems. However in Ref.[9], it does not consider the transmission time which is notneglectable from the RLP layer to MAC layer at the transmit ter side or from the MAC layer to RLP layer at the receiverside. So the analysis lacks integrity. For Ref. [10], al though it proposes an analytical model for RLP performanceconsidering all possible delay, it is still not to analyze thepacket error rate (PER) and packet transmission delay seenby upper layer above RLP layer. Therefore it does not provethe advantages of two layer retransmission compared with theRLP layer ARQ used in IS 99. In this paper, an analytical model of two layer retransmis sion is proposed, which considers all possible transmissiontime experienced by upper layer packets and analyzes thepacket error rate and transmission delay seen by upper layer(TCP in this case). By comparing the PER and transmissiondelay with conventional RLP layer ARQ, this model provesthat the two layer retransmission scheme can greatly decreasethe PER and transmission delay. The results also show thatthe soft packet combining in MAC layer can extremely in crease the correct probability of received frames. The rest of the paper is organized as follows. Section 2describes the structure of the two layer retransmissionscheme. Section 3 introduces the performance modeling andgives the expressions of PER and transmission delay seen byTCP. The analytical results are shown in section 4. Finallysection 5 concludes the paper.2 The Principle of Two Layer Retrans mission Scheme The structure of two layer retransmission scheme is shownin Fig.1. The data segment from upper lay

1 2 3 4

问答

我要提问