Research On Node Sleep Scheduling Approaches For Low-Duty-Cycle WSN

In recent years,with the rapid development of science,people have invented a variety of micro sensors with sensing,computing and communication capabilities.This is closely related to the continuous progress of distributed data processing technology and sensor technology,as well as the growing maturity of wireless communication technology and embedded technology.However,in Wireless Sensor Networks(WSN),the sensor node is limited in energy due to its limited size,and it is difficult to implement charging or battery replacement in reality.Therefore,it is necessary for the node to stay in the sleep state for a long time,in another word,to stay in the low-duty-cycle mode,for the low flow and energy limited WSN.However,the operation of low-duty-cycle also brings a big problem of sleep delaying to the network.Based on this,it is necessary to adopt a certain method of sleep scheduling to enhance the performance and prolong the lifetime of the network.In this paper,we primarily use adding the least times of wakeup to the node to reduce the end to end sleep delay at the expense of minimal additional energy consumption,and ensure the real-time transmission of data.The main work is as follows:(1)Based on the dynamic energy sensing,we have proposed a node sleep scheduling algorithm,which adaptively increases the number of wakeup timeslots for the problem of sleep delay in low-duty-cycle WSN.In the low-duty-cycle mode,each sensor node has its own scheduling table,allowing the node's wakeup time slot alternating with the sleep time slot.The algorithm performs sleep wakeup scheduling for the nodes in a link,determines the number of times to increase the wakeup time slots according to the perceived energy left in the node,and equalizes the energy load under the requirement of a given sleep delay.Experimental results show that,the algorithm reduces sleep delay and improves network performance.(2)On the basis of the mentioned sleep scheduling algorithm,we have broken the situation of given sleep delay and proposed a node sleep scheduling algorithm based on multi-wakeup mechanism.In this algorithm,the nodes don't need to be limited by the given delay.We can identify a lower bound for any topology and make the delay and energy consumption reach a balanced optimal value.Different from the previous single-wakeup mechanism,the algorithm preferentially selects the nodes with higher weights as Allocation Nodes(AN)to allocate new wakeup time slots by calculating the weight of the nodes on the link.Experimental results show that,the sleep scheduling method which uses multiple wake-up mechanism reduces the sleep delay on the link and decreases network energy consumption.(3)Aiming at the tradeoff between energy consumption and sleep delay in low-duty-cycle WSN,a sleep/awake scheduling algorithm based on Q-learning has been proposed.This algorithm in the method of Q-learning allows each node to make adaptions according to its current situation and the approximation of the neighbor situation,autonomously determining the state of each time slot.Since this approximation does not require communication with neighbors,the corresponding process energy consumption can be avoided.The comparison of experiments shows that,the algorithm extends the life cycle of the low-duty-cycle WSN.