Optimized query-driven appointment routing based on Expectation-Maximization in wireless sensor networks

  • Authors:
  • Hamid Shokrzadeh;Siavash Khorsandi;Abolfazl Toroghi Haghighat

  • Affiliations:
  • Islamic Azad University, Science and Research Branch, Tehran, Iran;Amirkabir University of Technology, Tehran 15875, Iran;Islamic Azad University, Qazvin Branch, Qazvin, Iran

  • Venue:
  • Journal of Network and Computer Applications
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Query-driven routing plays an important role in wireless sensor networks, especially when Sink looks for data with unknown location in the network. Rumor routing propagates event information in some select paths in the network, in effect, creating an event trace. When an agent on an event trace is crossed by a query agent, a route to event location is established. The main goal in all rumor-based algorithms is to expedite the cross over by the query-agents. This paper introduces a new class of rumor-based algorithms based on a concept called appointment, adding determinism to the cross-over point location as well as agent traversal trajectory in a randomly set-up network and event scenarios. Two flavors of the algorithm are proposed. In the first approach, query-agents and event-agents make an appointment at the edge of the network. In the second approach, the appointment is made at a central point in the network. We study the optimality conditions for such a central point as a function of the system parameters. A Gaussian mixture model is used to model the events' spatial stochastic distribution, and a dynamic Expectation-Maximization algorithm is developed to estimate its rank and parameters. A data fusion algorithm is then applied to calculate the appointment point. We formulate a trade-off among search speed, route length, and network life time. Simulation results in various scenarios are used to compare the proposed algorithms to other competing algorithms and also to the optimal solution. The results illustrate that algorithms proposed in this paper are nearly optimal and are superior compared to other query-based routing algorithms.