Buffer Map Message Compression Based on Relevant Window in P2P Streaming Media System

TL;DR

This paper introduces a theoretical framework and novel compression schemes for reducing redundant buffer-map data in P2P streaming systems, achieving significant size reductions and providing guidelines for practical algorithm development.

Contribution

It presents a new correlation-based compression approach for buffer-maps, with proven correctness and theoretical limits, improving efficiency over traditional methods.

Findings

Buffer-map sizes reduced by up to 95% with proposed schemes.

Theoretical limits of compression gain derived using probability and information theory.

Simulations on UUSee system demonstrate substantial size reductions.

Abstract

Popular peer to peer streaming media systems such as PPLive and UUSee rely on periodic buffer-map exchange between peers for proper operation. The buffer-map exchange contains redundant information which causes non-negligible overhead. In this paper we present a theoretical framework to study how the overhead can be lowered. Differentiating from the traditional data compression approach, we do not treat each buffer-map as an isolated data block, but consider the correlations between the sequentially exchanged buffer-maps. Under this framework, two buffer-map compression schemes are proposed and the correctness of the schemes is proved mathematically. Moreover, we derive the theoretical limit of compression gain based on probability theory and information theory. Based on the system parameters of UUSee (a popular P2P streaming platform), our simulations show that the buffer-map sizes are reduced by 86% and 90% (from 456 bits down to only 66 bits and 46 bits) respectively after applying our schemes. Furthermore, by combining with the traditional compression methods (on individual blocks), the sizes are decreased by 91% and 95% (to 42 bits and 24 bits) respectively. Our study provides a guideline for developing practical compression algorithms.