The focus of this paper is to analyze the relative sensitivity of the bulk
throughput performance of different versions of TCP, viz,, OldTahoe, Tahoe,
Reno, and New Reno, to channel errors that are correlated. We investigate
the performance of a single wireless TCP connection in a local environment
by modeling the correlated packet loss/error process (e.g,, as induced by a
multipath fading channel) as a first-order Markov chain. A major contribut
ion of the paper is a unified analytical approach which allows the evaluati
on of the throughput performance of various versions of TCP, The main findi
ngs of this study are that 1) error correlations significantly affect the p
erformance of TCP, and in particular may result in considerably better perf
ormance for Tahoe and NewReno; and 2) over slowly fading channels which are
characterized by significant channel memory, Tahoe performs as well as New
Reno. This leads us to conclude that a clever design of the lower layers th
at preserve error correlations, naturally present on wireless links because
of the fading behavior, could be an attractive alternative to the developm
ent or the use of more complex versions of TCP.