دانلود رایگان مقاله ICP: پروتکل خوشه بندی آنی برای شبکه های حسگر بی سیم

Abstract

Wireless sensor network (WSN) is one of the mainstay technologies in Internet of Things. In WSNs, clustering is to organize scattered sensor nodes into a cluster-topology network for communications. Existing efforts on clustering intensively focus on the energy-efficiency issue. However, in mission-critical applications, a fast clustering scheme, which can not only gather sensory data immediately after deployment but also reduce the energy consumption, is more desired. In this paper, we study the clustering problem considering both time- and energy-efficiency. We propose a novel instantaneous clustering protocol (ICP) that groups sensor nodes into single-hop clusters in a parallel manner. ICP can instantaneously complete the clustering due to two key designs. First, to determine the cluster heads locally. Existing methods require a long duration on cluster head voting. To waive the voting consumption, a cluster head in ICP is locally determined by the pre-assigned probability and its present status. Second, to minimize the amount of transmissions. Parallel transmissions from different cluster heads and acknowledgments (ACKs) from multiple cluster members lead to severe time and energy consumption. On the contrary, ICP gets rid of the ACK mechanism, instead, only cluster heads contend to broadcast during a given period. This period is elaborately derived to guarantee the connectivity. Experiments on a 64-node testbed and simulations on large-scale WSNs are extensively conducted to evaluate ICP. Performance results demonstrate that ICP significantly outperforms existing clustering methods by reducing up to 55% time consumption and 89% amount of transmissions for energy-saving.

8. Conclusion

Time and energy consumption are two fundamental metrics to evaluate the clustering in WSNs. In this paper, we present a parallel clustering method, namely ICP, to reduce both the time and energy consumption. The proposed ICP benefits from two key principles. First, the cluster heads are locally determined by the pre-assigned probability instead of voting. Second, the transmission load and the duration of clustering are minimized as long as the connectivity could be achieved. In this way, retransmissions and ACKs are removed. Moreover, ICP is a lightweight and fully distributed method. We implement ICP in the NetEye testbed using 64 TelosB nodes and conduct extensive simulations for large-scale WSNs. Results from experiments and simulations demonstrate that ICP significantly outperforms existing methods in terms of time and energy while ensuring the connectivity, load balance, and fault tolerance. ICP is promising in practical WSN applications, especially for mission-critical applications.