Delay tomography via compressed sensing under exponential distribution

Abstract

Network tomography allows the measurements of end-to-end to infer network internal links characteristics such as packet loss rates and delay. In this work, we apply concepts from compressed sensing and Maximum A-Posteriori (MAP) estimation to the delay estimation problem. We propose a new delay tomography scheme which can be considered as a variation of l 1 -l 2 optimization. The proposed scheme assumes the link delay is exponential which goes further than just considering sparsity of it in former works. We also consider the clock synchronization error between source and receiver in path delay measurement, which we treat as additive Gaussian random variables. We conduct a simulation performance analysis of delay estimation and congestion detection, demonstrating that higher estimation accuracy can be obtained through the proposed scheme.