| Internet fault diagnosis on time is important to network service providers, and Internet service providers (ISPs). For example, users always choose more reliable ISPs. With the diversity and heterogeneity, Internet fault diagnosis becomes an increasingly challenging problem.At early years, several router-based Internet diagnosis tools need the collaboration of the network nodes, these tools generally need special support from routers. These tools are unsecurity and reduce the network service. In recent years, scholars have proposed a tomography method. With the efficiency and accuracy of tomography, it has more and more researchers’attention. Tomography infer path level and link level properties through end-to-end measurements. However, the probes cause cross-traffic, links’properties cannot be uniquely determined from path measurements. So, we need an effective method based on tomography that can reduce cross-traffic and more accurately infer the performance of each path.In this paper, firstly, we propose a novel Path Loss Inference Algorithm (in short, PLIA). We selectively monitor linearly independent paths with a variant of QR decomposition to deduce the loss rate and latency of the rest paths through an algebraic approach. Then we define a Minimal Indentifiable Link Sequence (in short, MILS) as a link sequence of minimal length whose properties can be uniquely identified through end-to-end measurements. Last, we propose a novel Least-biased End-to-end Link Loss Inference (in short, LELLI) system for inferring link-level loss rate. We also design efficient algorithms to find all the MILSes and infer their loss rates for diagnosis.We build a network simulation platform to carry out our experiments in two different scenarios and the results from the experiments demonstrate that our approach increases accuracy in monitoring congested links compared to other representative schemas’. Especially in a router-level topology, our approach can deliver almost a 12.5% increase. |
PDF Full Text Request