Fast and upper bounded Fano decoding algorithm: queuing analysis

Abstract

AbstractIn computer networking, having limited buffer size is one of the vital causes of splitting the messages into smaller frames to be then transmitted over the network. Increasing the buffer size will not improve the case much but rather it is likely to degrade the performance of the system if it is not investigated properly. On the other hand, frames' retransmissions, as a result of frames being corrupted or dropped, need to be avoided as much as possible because of possible network congestion. To investigate the problem of undesired retransmissions, this paper addresses the use of convolutional codes utilising variable complexity decoding algorithms, such as the Fano algorithm, which mainly aim to mitigate frames' retransmissions caused by frames' corruption. It is noteworthy to mention that the rapid growth of wireless technologies has attracted everybody's attention. As wireless technology evolves, new decoding algorithms are emerging to demonstrate designs with variable power consumptions. Hence, in this paper, we propose a novel queuing model that describes not only frame arriving and departing processes after getting decoded according to how worse the channel condition is or what is the remaining system power but also a decoder with a double speed compared with what is best scrutinised in the literature. Moreover, in our buffering model, we can reduce the number of frames being dropped because of insufficient buffer size, and unwanted retransmissions owing to long frame waiting time, by virtue of bestowing larger buffer size, through our derived closed‐form expression that represents the expected number of awaiting frames in the buffers of Fano decoders parameterised by channel status, decoding threshold and frame arriving probability. As far as the system occupancy is concerned and over what have been previously proposed, our analytical results show impressive improvements. As such, through our proposed queuing model, the required decoder's buffer is substantially smaller of what that obtained by a prior model, considering the same networking parameters, which prompts decreasing the average frame waiting time. Copyright © 2015 John Wiley & Sons, Ltd.