A stochastic model of TCP/IP with stationary random losses

Abstract

Abstract-In this paper, we present a model for TCP/IP congestion control mechanism. The rate at which data is transmitted increases linearly in time until a packet loss is detected. At this point, the transmission rate is divided by a constant factor. Losses are generated by some exogenous random process which is assumed to be stationary ergodic. This allows us to account for any correlation and any distribution of inter-loss times. We obtain an explicit expression for the throughput of a TCP connection and bounds on the throughput when there is a limit on the window size. In addition, we study the effect of the Timeout mechanism on the throughput. A set of experiments is conducted over the real Internet and a comparison is provided with other models that make simple assumptions on the inter-loss time process. The comparison shows that our model approximates well the throughput of TCP for many distributions of inter-loss times.

Transport Protocols in the TCP Paradigm and their Performance

“…The result is that the process Z (t), if it starts out not very far from the new equilibrium value, "loses its memory" in a small multiple of 1 1−c units of time. We say that the process Z (·) has "relaxation time"…”

“…[25,59]): When a router recognizes one of its buffers is getting close to congestion, it can set a "Congestion Indicator Bit" in packets flowing through this buffer. To avoid confusion with the similarly named bit in ATM with ABR, Floyd and Ramakrishnan call this the "Congestion Experienced" (CE) bit, or, more accurately, the CE Codepoint (two bit-pattern) (1,1). In this note we use that name (CE), but make different use of the "CE bit pattern" and simply call it the CE bit.…”

Towards a Measurement Based Networking approach for Internet QoS improvement

“…3 The results presented in this paper have been obtained by analyzing Renater traffic, but the same results have been obtained on all traffics of all other analyzed networks. Results presented here then represent the actual current state of the art in traffic characterization.…”

“…Of course, the global throughput got using TFRC is not greater than the one using TCP, because TFRC aims at being TCP friendly, and then limits its sending rate according to the statistical TCP rate equations [3]. But, TFRC, and this is its second big advantage compared to TCP, induces less congestions and losses in the network, because of its reduced aggressiveness.…”

Interleaved Traffic Splitting: A promising technique to solve False Timeout

“…After showing that the use of variable pipe causes retransmission timeout even after TD is received by the sender, Sikdar et al modeled the performance of Sack under the bursty loss model in [7]. In [8], improved models for Sack performance under both bursty loss model and independent loss model are proposed by combining the work of [7,12,13]. In [6], Xiang et al proved that Sack shows the best performance among the different TCP implementations in OBS networks.…”