Search

Node localization obtained by estimating node positions is an essential technique for wireless multi-hop networks. In this paper, we present an optimized link state routing (OLSR)-based localization (ROULA) that satisfies the following key design requirements: (i) Independency from anchor nodes, (ii) robustness for non-convex network topology, and (iii) Compatibility with network protocol. ROULA is independent from anchor nodes and can obtain the correct node positions in non-convex network topology. In addition, ROULA is compatible with OLSR protocol, and it uses the inherent distance characteristic of multipoint relay (MPR) nodes. We reveal characteristics of MPR selection and the farthest 2-hop node selection used in ROULA, and describe how these node selections contribute to reducing the distance error for a localization scheme without ranging devices. We used a simulation to specify appropriate MPR_COVERAGE for ROULA, and give a comparative performance evaluation of ROULA for various scenarios including non-convex network topology and various deployment radii of anchor nodes. Our evaluation proves that ROULA is effective for localization in various network scenarios.

In an ad-hoc network where available bandwidth is limited, the amount of transmitted data needs to be reduced. Especially in mission-critical ad-hoc networks such as tactical or disaster rescue networks, the bandwidth constraints are much more severe than in commercial or general public networks. In order to decrease the data amount, flexibility of the transmitted information should be limited. In traditional information theory, the basic criterion for effectiveness of coding is entropy based on probability. In this paper, we propose a new measure, urgency index and urgency entropy, to evaluate the effectiveness of coding in urgent communication and discuss its usefulness for designs of bandwidth constrained, control-based ad-hoc networks from a viewpoint of utility functions and elasticity.

Determining the physical location of indoor objects is a key issue in ubiquitous computing. Although there have been many proposals to provide physical location, these have had several restrictions such as dependence on the type and size of objects and trade-offs between location accuracy and the number of sensing devices. We propose an approach to removing these restrictions using mobile detectors. This paper describes the design of a goods tracking system called MobiTra that estimates the locations of any indoor objects using detection histories sent by mobile detectors. MobiTra is separated by several components and two data flows. A prototype for MobiTra was developed using an active RF system and Cricket, which can obtain a high degree of accuracy. We evaluated how accurately MobiTra estimated location in an environment that generated MTR errors. The errors affected the size of the estimated area and its position. We propose a new method of estimating location to be used in environments that generate MTR errors.