A new distributed topology control algorithm based on optimization of delay and energy in wireless networks

Topology Control (TC) is one of the most important techniques used in wireless networks to obtain the desired network property. Most existing works with regard to TC focus on reducing energy consumption. Even though there are some works to consider delay in their resulting topologies, they do not consider the effect of radio interference on delay. Aiming at wireless sensor networks, we model a link delay as a function of the signal to interference noise ratio of the receiving node in this link and its packet forwarding time, and take a weight sum of delay and energy consumption as weight of edge (or link). The minimum weight sum of any edge can be solved by using the Get_min-cost_of_edge_(i,j)(i,j) algorithm proposed in this paper. An Optimal Edge-cost Topology Control (OETC) algorithm is proposed to ensure that all approximate minimum-edge-cost paths exist in final topology. We also propose a Distributed Symmetric Link Maintenance (DSLM) algorithm to ensure that all links are symmetric in final topology if all links in original topology are symmetric. We prove that the communication complexity and computational complexity in OETC+DLSM are O(Nu)O(Nu) and O(Ne∗Nu2) respectively, where NuNu denotes the number of any node uu’s neighbors and NeNe denotes the times of executing the Get_min-cost_of_edge_(i,j)(i,j) algorithm. Furthermore, we verify through simulation that the network topologies produced by OETC+DLSM show good performance in terms of expected average link delay and node hop-count while keeping average energy consumption at an acceptable level.

► We present a new delay metric which considers interference and resource of nodes.
► We model a multi-objective optimization function for delay and energy cost.
► Based on the optimization result, we design a new topology control algorithm.
► Our algorithm has very low delay and its energy cost is at an acceptable level.
► Our algorithm has less communication cost but slightly more computation overhead.