Integrated Packet Loss and Error Control Schemes in Wireless Multimedia Data Links Jinsong Wu Jacek Ilow CDMA BTS Algorithm Development Electrical and Computer Engineering Nortel Networks, Calgary, AB Dalhousie University, Halifax, NS Canada Canada wujs@ieee.org j.ilow@dal.ca Abstract - This paper describes a novel Forward Error Correction (FEC) based technique, VHPG FEC, which combats packet losses and packet errors simultaneously. The scheme proposed uses a two- dimensional (2-D) coding approach in which, at the encoder, information carrying packets are arranged vertically to form the 2-D matrix. The first code is applied along the horizontal direction to form the redundancy packets mainly to recover from lost packets; the second code is applied along the vertical direction within the individual packets to minimize the bit error rate (BER). Within the group of packets, the overall effect of this scheme is the reduction in the effective packet loss and the BER without need for retransmission. The scheme is applicable in environments where the packet loss is not exclusively a result of high BER and where the transmission time constraints are critical. The adaptive extension of the scheme proposed, AVHPG FEC, is introduced to deal with dynamic wireless channel characteristics. Performance of the VHPG coding is analyzed through theoretical calculations to illustrate the choice of coding parameters for various channels. 1. INTRODUCTION 1.1 Background Traditional FEC and ARQ error control schemes [1], which deal with packet level errors, are not efficient enough to confront the wireless communications environments with high BER and packet loss to meet guaranteed Quality of Service (QoS) in multimedia applications with stringent delay constraints. Another common error control method for combating burst bit error within packets is interleaving, but it is not advantageous when the inter-packet level has high BER and packet loss. Many existing solutions for providing QoS in wireless networks, such as [2] and [3], are based on the higher network layers. In the context of multimedia packet networks, limited adaptive wireless data link solutions, such as [4], have been explored. The focus of this paper is to meet the requirements of bounded delay and jitter in real-time wireless applications at the data link layer. The approach presented relies on a relatively low rate FEC code, and, with the multimedia traffic considered, the proposed solution is particularly applicable to new generations of broadband wireless access networks. 1.2 Problem formulation The critical observation that packet loss is not exclusively caused by erroneous transmission constitutes the foundation for the work described in this paper. Hard handoff, the process of buffer overflow, and network congestion are several possible reasons for packet loss independent of the BER level within the packets. The construction of redundancy packets using FEC with erasure decoding to decrease the packet loss and to ensure the fixed transmission delay and jitter has been proposed at layers higher than data link in [5] where erasures correspond to lost packets. All the previous work in this area assumes that once the packet (either redundancy or information) is received, it has no or very few errors in higher network layers. However, such an assumption requires an extremely powerful FEC code at the physical layer or ARQ at the lower layers. The first option, combined with a redundancy level packet, could introduce prohibitive overhead, while the second option could probably recover lost packets itself. Considering the drawbacks of conventional error-control schemes for wireless multimedia applications and the flaws of the newer FEC-based packet loss control schemes, this paper presents a novel approach at the lower layers, especially the data link layer, to combat packet errors and loss at the same time, using an integrated FEC method. The breakthrough of the proposed method comes from applying FEC along two dimensions: first, across the information packets to construct the redundancy packets, and second, within individual packets to protect them against erroneous transmission. The distinguishing feature of the coding scheme proposed is that, unlike in conventional 2-D codes, the columns of the 2-D code (packets) are transmitted as independent entities. 2. VHPG FEC 2.1 Basic ideas The 2-D FEC scheme presented here, Vertical-Horizontal Packet Group FEC (VHPG FEC) [6], flexibly utilizes the product of conventional 1-D Reed-Solomon (R-S) codes in vertical and horizontal directions because of their superior performance in the presence of burst errors such as those in wireless links. At the receiver, the 1-D R-S decoding with hard decisions is performed first within the packet to reduce the BER and then within the group of packets 1-D erasure R-S decoding is employed to possibly recover from lost (“erased”) packets. The VHGP FEC coding scheme is characterized by the number of parameters that could be adapted to varying wireless channel conditions so that the throughput and the reliability of the data could be balanced. 2.2 Structure and possible implementation Fig. 1 shows the group of packets coding matrix. The packets belonging to a specific group are arranged column-wise into a matrix. In the figure, VH(m,n) stands for the m th bit of the n th packet. The core idea in the proposed VHPG FEC code is that coding is applied to a group of numbered packets, and the packets 0-7803-7206-9/01/$17.00 © 2001 IEEE 685