Efficiency of navigation in indexed networks

TL;DR

This paper explores efficient packet navigation methods in complex, indexed networks, demonstrating a scheme that nearly reaches the theoretical maximum efficiency across various network models.

Contribution

It introduces and tests index-based navigation schemes in complex networks, identifying a highly effective method close to optimal performance.

Findings

One scheme outperforms others in efficiency.

The best scheme approaches the theoretical optimum.

Effective navigation schemes are applicable to distributed systems.

Abstract

We investigate efficient methods for packets to navigate in complex networks. The packets are assumed to have memory, but no previous knowledge of the graph. We assume the graph to be indexed, i.e. every vertex is associated with a number (accessible to the packets) between one and the size of the graph. We test different schemes to assign indices and utilize them in packet navigation. Four different network models with very different topological characteristics are used for testing the schemes. We find that one scheme outperform the others, and has an efficiency close to the theoretical optimum. We discuss the use of indexed-graph navigation in peer-to-peer networking and other distributed information systems.