Walter Featherstone

N-in-1 Retransmission with Network Coding

Retransmission (ReTX) mechanisms, i.e. Automatic Retransmission reQuest (ARQ) protocols are widely used for error control in mobile communications. Using ARQ, corrupted data frames are retransmitted in several attempts for eventual error recovery at the receiver. Error correction coding and soft combining have also been used in Hybrid ARQ (HARQ) protocols to increase efficiency. However, the evolution of ARQ protocols to date has been limited to the paradigm of ONE ReTX attempt for (the recovery of) ONE data frame, which often leads to excessive redundancies. In this paper we propose a new N-in-1 ReTX scheme. Using Random Network Coding, the scheme further improves the efficiency of ReTX by transforming a data frame into a series of equally-useful blocks for error recovery and sharing every ReTX attempt among multiple frames. Results from extensive simulations show that the proposed scheme achieves a significant throughput gain of up to 106% against the conventional HARQ-CC protocol.

System level performance evaluation of GPRS for various traffic models

During the GSM standardisation process data services were provisioned for. However, these were based on circuit switched connection. It is now widely accepted that circuit switched connection does not provide optimum utilisation of a communication channel for data transfer. The provision of more efficient radio data bearers has become an essential piece of the existing technologies such as GSM and the future ones such as UMTS. In GSM, ETSI finalised a new packet data service called GPRS in mid 1998. The GPRS data rates will provide bit pipes in the range 8-20 kbps per time slot. GPRS will be deployed in existing GSM networks that have been optimised for voice (circuit switched data) traffic. This paper investigates the system level performance of the GPRS coding schemes in such networks using a system level simulator. The performance is quantified in terms of the throughput and delay for various types of traffic. The results can be used to generate optimisation rules and efficient resource management algorithms.

Performance analysis of varying the link adaptation BLER switching point in EGPRS

This paper examines the impact of the choice of the link adaptation BLER switching point on the performance of Enhanced General Packet Radio Service (EGPRS) Phase I, with the aim of determining the optimum switching point. The analysis is carried out by considering outputs from a system level simulator and a system dimensioning and planning tool. Results are presented for both LA and IR for a range of LA switching points. A GSM macrocellular environment, having an interference-limited 4/spl times/3 re-use pattern, was chosen. The mobiles operate with perfect receivers in a variety of radio channel conditions.

System level performance evaluation of EDGE Compact

EDGE Compact has been developed to enable the deployment of an EGPRS system by utilising only 0.6 MHz worth of spectrum. The implication of using such a narrow spectrum is that system loading becomes a critical factor in maintaining an acceptable QoS service from the users perspective. This paper investigates - under a simulation environment - the performance of 1:3, 2:6, 3:9 and 4:12 EDGE Compact. The scenarios are analysed by varying system loading and considering system throughput and user delay statistics with and without admission control. The results are particular to the population of mobiles operating with perfect receiver, under TU50-NoFH radio condition, E-mail traffic and perfect link adaptation.