On the Impact of Random Losses on TCP Performance in Coded Wireless Mesh Networks

Abstract

Network coding (NC) is a promising technique to improve throughput in wireless mesh networks. However, some previous studies have found that the actual performance improvements offered by NC may be much lower than the gains predicted by theory. This is especially so when the bulk of the traffic carried by the mesh network is composed of TCP flows. By means of both mathematical modeling and ns-2 simulations, we explore some issues due to random packet loss in coded mesh networks. Our results illustrate how the use of network coding may induce synchronization between TCP flows; they also suggest that, under particular conditions of random packet loss, the aggregate throughput may actually be lower when NC is used.

TCP with hop-oriented network coding in multi-radio multi-channel wireless mesh networks

“…Secondly, the throughput gain derived by NC approaches varies as network topologies, traffic patterns and packet loss rates change. Thirdly, throughput gains achieved by NC may be remarkably lower than expected in practical application [9], in terms of TCP flows, because the TCP congestion control algorithm is not well compatible with the behaviour of NC. In addition, merging network coding and TCP together may increase the end-to-end delay.…”

“…(5) else if b is an ACK-NCB packet whose block index is j, (a) if r j = L mk , B ′′ mk − −, recalculate p mk , d mk , recalculate B ′ mk , B mk and C f m according to (9), (11) and (18), (b) else, re-send ⌊R mk L mk ⌋ 2 r j NC packets of block j Recalculate p mk , d mk : p mk and d mk are estimated by the way that RTT in TCP does, as shown in (18), (19). P mk_this is equal to 1 2 r mk /s mk and d mk_this is the d mk value obtained in an ACK-NCB currently.…”

“…The reasons can be briefly stated. According to (9), the estimated propagation delay is in direct proportion to the number of available codecs B ′ mn . A small B ′ mn means there may not be enough codecs to fully utilise the bandwidth.…”

A Model for Steady State Throughput of TCP CUBIC

“…We assume that random packet loss experiences a random Poisson process with rate λ. This assumption has also been made in [15]. Given a time instant t 0 , the time duration τ loss from time t 0 to the next loss event is a random variable with an exponential distribution.…”

“…The complex, nonlinear window growth function of TCP CUBIC seems to be the bottleneck of establishing an appropriate model for TCP CUBIC. The fluid model in [15] took both random packet loss and congestion loss into consideration, but it is valid only for additive increase multiplicative decrease (AMID) cases. In [16], an increase rate accelerator (AIRA) model was proposed.…”