Modified Strategy to Improve QoS in Networks with Varied Traffics.

Abstract

The efficiency of buffers in network routers plays an important role in effectively accommodating packets that arrive in bursts at the routers interfaces. The choice of a suitable router buffer size is still a significant problem. Since, the use of small buffer guarantees low packet delay but higher chances of packet lost. Similarly, larger buffer leads to buffer bloat which causes higher delay in a network resulting to poor Quality of Service (QoS). Buffer bloat is a significant problem due to the high changing link characteristics of modern heterogeneous network traffics. The access links can have connections with speedy links (Gbps) with small amount of packet losses and have connection with a susceptible high packet loss with low-bandwidth links like wireless and last mile connections. Hence, what may be thought to be a rational buffer size might be flawed when link rates and delay fall below the minimum value. Thus, this paper analyses buffer size performance and optimization in networks with heterogeneous traffics for Random Early Detection (RED) Queues using NS2 simulations to obtain a range of better suited buffer sizes that improve the QoS without extra signaling and computation by routers. The results obtained established that having a buffer size between five to ten kilobytes yielded best average throughput with low average delay for RED Queue at the congested router Interface. In addition, the optimized buffer size scheme improved the QoS accordingly. Keywords: Buffer Size, Network Traffic, Quality of Service, Simulation