Research On Sleep Scheduling Algorithm For Wireless Sensor Networks

Wireless sensor networks (WSNs) have the characteristics such as hardware resource limitation, power capacity limitation, data-central, dense distribution with numerous nodes etc, so how to balance energy consumption of the network to maximize the network's lifetime should be put on the first place when realizing the key technologies of WSNs.At present, node scheduling algorithm is one of the efficient methods to extend the lifetime of WSNs, however, it does not apply to wireless sensor network applications with high accuracy, because the sleeping of nodes will cause data missing and reduce accuracy. Therefore, combined with data prediction techniques, an energy-balanced sleep scheduling mechanism is proposed first, which saves node energy and also enables energy consumption to be distributed evenly throughout the whole network, and delays the node death time and we also make theoretical analysis by formula derivation. Next, a data temporal-correlation-based prediction algorithm is proposed, by adopting unbiased GM (1, 1) model, it uses the history data to estimate the data in future time; Secondly, a data spatial-correlation-based prediction algorithm is proposed, by using grey correlation analysis method to analysis the data between nodes, combined with the data of unsleeping nodes in the same cluster, it predicts data of sleeping node indirectly. Based on the above two algorithms, an adaptive data temporal-and-spatial-based prediction algorithm is proposed, based on information maximization principle, and taking the minimum sum of error absolute as the target function to build combination forecasting model, and overcomes the shortcomings of the single model, which also reduces the predictive randomness.Theoretical analysis and simulations with real data indicate that the proposed algorithms can make a good performance on balancing node energy consumption and extending the network lifetime, and also ensuring high data accuracy.