Performance Evaluation of Error Control Protocols over Finite-State Markovian Channels Sohraab Soltani † Hayder Radha ‡ ∗ † Department of Computer Science & Engineering, ‡ Department of Electrical & Computer Engineering Michigan State University {soltanis, radha}@msu.edu Abstract— Reliable transmission is a challenging task over wireless links because wireless channels are susceptible to noise and interfer- ence. To address this problem various error control schemes have been introduced which are broadly categorized into automatic repeat request (ARQ)-based and forward error correction (FEC)- based techniques. In [7], we introduced a channel adaptive error control protocol (PEEC) which utilized a buffering mechanism and the receiver feedback for optimal error recovery. In this paper, we are interested in evaluating the performances of ARQ- , FEC-based error control schemes and PEEC protocol over a Finite-State Markov Channel (FSMC). I. I NTRODUCTION Reliable transmissions over wireless channels are challeng- ing because wireless links are error-prone and susceptible to noise imposed by fading, interference and mobility. Therefore the current IEEE 802.11 protocol is designed to be as reliable as possible [1], incorporating frame check sequence (FCS) for detecting errors and automatic repeat request (ARQ) to retrans- mit corrupted packets. The IEEE 802.11 link layer discards corrupted packets without regard to the number and location of the errors. This approach is suitable for wireless channels with relatively low bit error rates (BER) because the likelihood of receiving consecutive corrupted packets is small and the original packet can be delivered after few retransmissions. However, for channels with more severe error (and arguably more realistic conditions), the IEEE standard ARQ scheme can cause multiple retransmissions even if there is a single error in a packet. This in turn leads to the transmission of a large number of redundant (correct) data. As a result, the overall throughput deteriorates steadily and rapidly with increasing average channel BER. To enhance the IEEE standard ARQ performance, packet combining techniques have been developed in which they exploit multiple transmissions typical of ARQ schemes. The basic idea is to store previous transmissions of the corrupted copies of a packet and attempt error recovery. Techniques developed in this context are xor combining [2] and majority combining schemes [3] [4]. The advantage is that they are IEEE802.11 MAC compatible; however it has been shown that the improvement of the throughput is not remarkable [6]. * This work was supported in part by NSF Award CNS-0721550, NSF Award CCF 0728996, NSF Award CCF-0515253, and NSF Award CNS-0430436. In recent years, many papers in multimedia applications have proposed cross-layer mechanisms to overcome performance limitations imposed by conventional protocols. The analysis of the hybrid Erasure-Error protocols (HEEPs) in [5] shows that cross-layer protocols in general provide capacity im- provement in many realistic scenarios and that they signifi- cantly improve the overall performance as measured by video quality. However, the drawback of the cross-layer protocols is that their implementations require major modifications in transport and application layers. Another alternative includes well established hybrid ARQ (HARQ) schemes [8] [9], where the combination of ARQ and forward error correction (FEC) can be used. These schemes introduce additional complexity which stems from FEC operations but at the same time improves the overall throughput over channels with high BER. Although these schemes perform better in noisy channels, their performances are below the IEEE standard ARQ for channels with good conditions. In [7], we introduced an error-combating scheme at a link layer, which we refer to as Packet Embedded Error Con- trol (PEEC) protocol. PEEC employs packet-embedded parity symbols (instead of retransmission) for error recovery. Further, PEEC utilizes acknowledgment flags to asses the channel and the receiver buffer conditions. Depending on this assessment, PEEC adaptively administers the transmission of the data and redundancy (parity) symbols such that the level of parity symbols guarantee high likelihood of successful recovery of new data as well as corrupted data at the receiver while the available bandwidth is efficiently utilized for the transmission of new data. The performance analysis in [7] indicates that PEEC achieves higher throughput than other error-combating schemes, regardless of channel condition. In this paper, we evaluate the performances (as measured by decoding failure likelihood) of ARQ-, FEC-based error con- trol schemes and PEEC protocol over a Finite-State Markov Channel (FSMC). Using the FSMC model, we first find the bounds of channel’s Shannon capacity based on the amount of available information provided in receiver’s feedback. Sec- ond, we explore the functionality of various error correction schemes and derive the likelihood of decoding failure of each method. Finally, to determine whether any of the proposed error control mechanisms perform optimally with respect to