Deadlock Avoidance Policy in Mobile Wireless Sensor Networks with Free Choice Resource Routing

Efficient control schemes are required for effective cooperation of robot teams in a mobile wireless sensor network. If the robots (resources) are also in charge of executing multiple simultaneous missions, then risks of deadlocks due to the presence of shared resources among different missions incr...

Full description

Bibliographic Details
Main Authors: Prasanna Ballal, Abhishek Trivedi, Frank Lewis
Format: Article
Language:English
Published: SAGE Publishing 2008-09-01
Series:International Journal of Advanced Robotic Systems
Online Access:https://doi.org/10.5772/5607
Description
Summary:Efficient control schemes are required for effective cooperation of robot teams in a mobile wireless sensor network. If the robots (resources) are also in charge of executing multiple simultaneous missions, then risks of deadlocks due to the presence of shared resources among different missions increase and have to be tackled. Discrete event control with deadlock avoidance has been used in the past for robot team coordination for the case of multi reentrant flowline models with shared resources. In this paper we present an analysis of deadlock avoidance for a generalized case of multi reentrant flow line systems (MRF) called the Free Choice Multi Reentrant Flow Line systems (FMRF). In FMRF, some tasks have multiple resource choices; hence routing decisions have to be made and current results in deadlock avoidance for MRF do not hold. This analysis is based on the so-called Circular Waits (CW) of the resources in the system. For FMRF, the well known notions of Critical Siphons and Critical Subsystems must be generalized and we redefine these objects for such systems. Our second contribution provides a matrix formulation that efficiently computes the objects required for deadlock avoidance in FMRF systems. A MAXWIP dispatching policy is formulated for deadlock avoidance in FMRF systems. According to this policy, deadlock in FMRF is avoided by limiting the work in progress (WIP) in the critical subsystems of each CW. Implemented results of the proposed scheme in a WSN test-bed is presented in the paper.
ISSN:1729-8814