A Frequency-Time Synchronization Scheme for Real-Time Wireless Sensor Networks

Abstract

Listen

Translate article

In a real-time wireless sensor network (RT-WSN), an unpredictable time length of the synchronization (or connection) process between nodes is generally a pity, though the communication after the connection may be controllable. The purpose of this paper is to solve this kind of pity based on the multiple-request-single occasion (multiple slave nodes request to send data to a single master node simultaneously before getting synchronized using the frequency channel hopping technique). Suppose that the master sends the synchronization packet (or beacon) and the slaves scan for this packet with different channels for connection. A slave getting synchronized with the master means that both nodes have just selected an identical frequency channel during a time region and the slave has received the synchronization packet successfully in this region, which is called frequency and time synchronization, abbreviated as FTS. For many existing wireless protocols, if they are directly adopted in this situation, two deficiencies exist as for real-time performances: First, the time length required for a slave to join the network is often not deterministic if one or more channels are disturbed. Second, when multiple slaves do their scanning simultaneously, which slave can synchronize with the master first is unpredictable so that a slave with a lower priority may be serviced prior to others. In this paper, two FTS examples with poor real-time performances are provided first. Then, a synchronization method named 1/2n FTS is presented and proved. With this method, a slave scans for the synchronization packet of the master with n different available channels repeatedly until it gets the packet while the master transmits the packet 2n times in 2n continuous timeslots. The width of the scan widow of the salve takes twice the width of the slot. In this way, every slave has the opportunity to get synchronized with the master at the end of the 2n slots even if one or more (not all) channels are disturbed. Then, the slaves can send their requests to the master in different slots so that the master can schedule subsequent communications according to their requests and priorities. Also, if the mater broadcasts the beacon periodically, the time length for a slave to join or rejoin the master is not difficult to predict. The theorems associated with the 1/2n FTS method are demonstrated in experiments with NORDIC Semiconductor chips.