An adaptive medium access control scheme for mobile ad hoc networks under self-similar traffic

Abu-Tair, Mamun; Min, Geyong; Ni, Qiang; Liu, Hong · 2009 · Crossref

DOI: 10.1007/s11227-009-0324-3

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the inadequacy of traditional traffic models in capturing the complex, long-range dependent (self-similar) characteristics of modern network traffic, which arises from multiplexed data, voice, and video streams. The authors aim to analyze the impact of these traffic patterns on Quality of Service (QoS) in packet networks, specifically focusing on packet loss behavior. Recognizing that simple loss statistics fail to capture the complexity of loss patterns, the study characterizes packet loss using "loss episodes" to reflect the burstiness inherent in real-time applications, which are particularly susceptible to consecutive packet losses. The research employs trace-driven network simulations using the ns-2 simulator. MPEG-1 traffic traces from various sources, including movies and news programs, are transmitted over UDP/IP to simulate real-time applications. The simulation topology involves a router with five different queuing schemes: FIFO/DropTail, Random Early Drop (RED), Fair Queuing (FQ), Stochastic Fair Queuing (SFQ), and Deficit Round Robin (DRR). Router buffer sizes are set according to delay requirements, tested at 46, 100, and 200 packets, with a packet size of 552 bytes. The study utilizes wavelet analysis to examine both the input traffic and the resulting packet loss traces. The results demonstrate that the input traffic traces exhibit long-range dependency (LRD) for time scales of approximately 1 second. Crucially, the analysis reveals that the packet loss traces also exhibit LRD for time scales of approximately 1.2 seconds, indicating that the loss process captures the self-similar characteristics of the underlying traffic. This LRD behavior in loss processes is present regardless of the buffer size or the specific queuing scheme employed. The study further distinguishes between aggregate loss and per-flow loss, showing how loss episodes of varying lengths contribute to the overall loss pattern. For instance, aggregate loss may consist of multiple episodes, while individual flows experience distinct loss patterns. The significance of these findings lies in the confirmation that self-similarity is not only a property of the input traffic but also of the resulting packet loss processes. This implies that traditional models assuming independent or short-range dependent loss are insufficient for accurately predicting QoS in networks handling self-similar traffic. The persistence of LRD in loss traces across different buffer sizes and queuing disciplines suggests that network designers must account for these long-range dependencies when configuring routers and designing QoS mechanisms for real-time applications. The use of wavelet analysis provides a robust method for identifying and quantifying these self-similar properties in both traffic and loss patterns.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• traffic density