Performance of TCP congestion control with explicit rate feedback: rate adaptive TCP (RATCP)

We consider a modification of TCP congestion control in which the congestion window is adapted to explicit bottleneck rate feedback; we call this RATCP (rate adaptive TCP). Our goal in this paper is to study the performance of RATCP (using analytical models, and an experimental test-bed) in various network scenarios, and to compare the performance of RATCP and TCP, both with and without fast-retransmit and fast-recovery. We find that when sessions with the same round trip times share a bottleneck link then, even with ideal fair rate feedback, the performance of RATCP is only slightly better than that of TCP. RATCP, however, does reduce losses significantly. When there are random losses, however, RATCP with fast-recovery provides substantially better throughput than plain TCP. Further, when sessions with different round trip times share the bottleneck link, as expected, RATCP ensures fairness. Finally, we suggest a practical situation in which RATCP can be useful for improving Web access performance