IEEE1588Time Synchronization Optimization And OMNeT Simulation For WSNs

With the rapid development and mature of wireless communication technologies, especially the advantages of wireless transmission and easy-to-fast networking, application fields of Wireless Sensor Networks (WSNs) continue to expand. In many new applications, keeping high consistency in clocks of distributed network nodes are needed, in order to achieve more accurate data sampling to improve the quality of the acquired data, and also to better schedule data transmission, reduce power consumption, extend network lifespan. Therefore, as an emerging research field of WSNs, time synchronization technology for WSNs has been more and more attention in recent years, especially in time-sensitive distributed applications, time synchronization in WSNs has become a critical core technology.Although many time synchronization algorithms and protocols have been proposed for packet-exchange networks, such as the most widely accepted Network Time Protocol (NTP) in the Internet, the recently proposed IEEE1588Precise Time Protocol (PTP), and Flooding Time Synchronization Protocol (FTSP). etc. NTP has lower accuracy, usually in tens of milliseconds, FTSP has higher requirements on communication bandwidth, while PTP proposed as the IEEE1588standard is designed for high precision time synchronization, is able to obtain high synchronization accuracy by a relatively low-bandwidth communication, and is expected to become the future standard for WSNs time synchronization. However, PTP is designed for wired industrial Ethernet, and has high requirements on the symmetry of propagation delay and the accuracy of time-stamps. But the wireless data transmission is usually affected by channel sharing, wireless channel fading, node mobility, packet collisions and retransmissions and other factors, resulting in larger transmission delay jitters, it is difficult to meet the requirements of PTP. Meanwhile, in low-cost WSNs, due to the limited resources, it is difficult to achieve accurate time-stamps. Therefore, delay jitter, timestamping uncertainties and other factors’effects on PTP are needed to in-depthly study, so as to provide guidance for improving PTP algorithm. However, due to the CSMA/CA media access control protocol commonly used in WSNs has higher complexity and randomness, the relationship between delay jitters caused the CSMA/CA and the time synchronization accuracy usually can only obtained through a lot of simulations.The main challenges in PTP time synchronization for WSNs are:1) According to the characteristics of WSNs, how to build mathematical model of crystal clocks and PTP time synchronization protocol, build a more realistic software simulation platform for high precision simulation of delay jitter, timestamping uncertainty and synchronization accuracy to reveal the relationship among them, and verify the performance of time synchronization algorithm by a large number of simulations.2) Combining closed-loop feedback control technique in the modern control theory and state estimation technique with time synchronization, so as to study how to improve and enhance the PTP synchronization algorithm. These are also the main tasks of our study presented in this thesis. More specifically, the main content and features of this thesis are:(1) Understanding the nature of time synchronization. Time synchronization is a process of estimating the time difference between a drifting local clock and the global reference time and adjusting the local clock to keep the difference in a smaller range. Time synchronization is essentially a process to filter noises from a series of time measurements with noises and accurately estimate the clock offset. For computer network based on packet exchange, most clock synchronization algorithms are through exchanging the special time-stamped packets to achieve clock offset estimate, therefore, time synchronization is also a process for time accuracy at the expense of communication bandwidth and hardware and software consumption.(2) State space modelling of WSN’s oscillator clocks and the PTP protocol. The main task in this part focuses on detailedly analyzing the discrete process of the continuous-time model of the physical clock to build discrete-time mathematical models of oscillator clocks and the PTP protocol. On the other hand, based on the understanding of the essence of time synchronization, when building the clock model and PTP protocol model, the clock model is converted to a state transition equation and the PTP protocol model is converted to an observation equation, so as to innovatively build a state-space model for time synchronization. Thus lays a foundation for employing closed-loop feedback control technique and state estimation technique, etc.(3) Building a discrete event simulator TS2(Time Synchronization Simulator) based on OMNeT++, and implementing a high realistic simulation for PTP time synchronization in WSN based on the IEEE802.15.4(TI CC2420). The simulator has the features of high reality, easy to extend and reuse modules, supporting mobile nodes and large-scale network.(4) On the basis of the state-space model, two improved algorithms for PTP time synchronization are proposed, respectively are servo clock algorithms based on proportional controller and Kalman filter. Simulation results show that:compared to only PTP synchronization algorithm, PTP synchronization algorithm based on the proportional controller and the Kalman filter optimization exhibit significant advantages on both synchronization accuracy and stability.