Level the buffer wall: Fair channel assignment in wireless sensor networks

  • Authors:
  • Yanyan Yang;Yunhuai Liu;Lionel M. Ni

  • Affiliations:
  • Department of Computer Science and Engineering, Hong Kong University of Science and Technology, Hong Kong;Department of Computer Science and Engineering, Hong Kong University of Science and Technology, Hong Kong;Department of Computer Science and Engineering, Hong Kong University of Science and Technology, Hong Kong

  • Venue:
  • Computer Communications
  • Year:
  • 2010

Quantified Score

Hi-index 0.24

Visualization

Abstract

In this paper, we study the trade-off between network throughput and fairness in a multi-channel enabled wireless sensor network (WSN). Traditional approaches attempt to solve the two problems in an isolated manner without a joint design. Although there are schemes aiming to improve either throughput or fairness, respectively, our empirical studies show that solutions to these two problems cannot be simply combined to achieve better throughput and fairness simultaneously. Away from the traditional belief, the number of channels in WSNs with Telosb sensor nodes operating at 2.4GHz band can be up to 83 and the orthogonal channels can be up to 27. The switching overhead in terms of time and energy cost is relatively small. The averaged switching time for a sender is about 0.34ms and the additional energy consumption caused by channel switching is no more than 2% when 10^4 packets with 43bytes of each are sending out continuously. Furthermore, we observe a buffer wall phenomenon which is one of the main reasons causing network throughput degradation and unfairness. To strike a better trade-off between the network throughput and fairness, we design a novel multi-channel assignment algorithm, targeting at maximizing the minimal data sending rate. The key idea of the proposed algorithm is to level down the buffer wall so that the buffer usage of nodes can be evenly distributed. As such, the bandwidth of bottleneck nodes can be fully utilized and the unfairness due to the node locality can be removed. We prove that the achieved data sending rate is no less than 4/9 of the optimal rate in theory. Our experimental results based on both chain and uniformly random sensor deployment topologies show that the minimal data sending rate can be improved by up to 100% comparing with the existing work TMCP.