Balanced clustering approach to enhance lifetime and throughput of wireless sensor networks

Abstract

Wireless sensor networks (WSN) is a special type of Micro Electro-Mechanical System (MEMS) which is composed of large number of small, inexpensive and low powered sensor nodes. Lifetime is one of the crucial challenges of WSN. Balanced clustering may be a suitable solution of this challenge. Proper selection of cluster head (CH), cluster formation and a suitable intra-cluster communication technique can create a balanced clustering. Balanced clustering will extend lifetime as well as throughput. Many protocol exist for clustering. They used various matrices like residual energy, position or distance from base station (BS), neighbor set, neighbor information etc., but none of them can create balanced clustering. Moreover, many of them create back transmission which consume further energy and degrade lifetime. In this research, we offered a balanced clustering by selecting CH, cluster formation and a suitable intra-cluster communication technique.

The parameters residual energy (RE), number of neighbor nodes (NNN), one-hop neighbor information (ONI) and distance from nodes to BS (DNB) have been used for selecting a CH. RE information helps to select comparatively higher energetic node as CH, NNN helps to select CH from better density area of nodes of the network, ONI will restrict to select one CH from one cluster and DNB will reduce back transmission path. In cluster formation, energy may be wasted due to too high or too low cluster size. We have restricted it by using central border, internal and balancing nodes. We have used two threshold value (maximum and minimum) for ensuring suitable cluster size. Most of the clustering approach used TDMA for intra-cluster communication. But huge TDMA time slots may be unused due to data un-availability or lower trafic of slots owner’s node. We have used three steps in intra-cluster communication (ICC) technique for sending more packets. At first, we used “power level adjustment” for sending non-owner node’s data to CH by exploiting capture effect. Secondly, we used “time slot adjustment” by adjusting window size which ensured whether a non-owner node should send data or not. Lastly, we used “Preamble based CSMA” with waiting time adjusting according to power level of data for sensing channel, it will reduce collision. These three steps, ensure more packets sending to CH and BS, hence throughput increased more and more compare to previous works.

The proposed methods are evaluated by OMNeT++ simulator and compared with LEACH-C, LEACH-MAC and an energy efficient and balanced clustering approach for improving throughput (EEBCAIT). It is found that major improvement of performance in terms of First Node Death (FND), Tenth Node Death (TND), End Node Death (END), Consumption of Energy vs Rounds. Remaining Energy vs Rounds, Alive Nodes vs Rounds, Dead Nodes vs Rounds, CHs vs Rounds, Total Packets vs Rounds, Packets sending per round and improvement using of idle slots per round. We also found great improvement of lifetime and throughput.