Dynamic and Quantitative Method of Analyzing Clock Inconsistency Factors among Distributed Nodes

Abstract

Designing appropriate clock synchronization schemes and maintaining time consistency among many sensor nodes are crucial for novel distributed applications environments such as Internet of Thing (IoT). However, analyzing the various factors leading to clock inconsistency should be conducted first. The traditional analysis methods are primarily experiential and qualitative, and dynamic disturbances existing among the factors are not considered; moreover, the emerging IoT is rapidly evolving in terms of large-scale feature, service-oriented trend, complexity, and dynamics. Such developments present difficulties in the use of traditional methods in IoT for the analysis of factorial effects on system clocks. To remedy these problems, we propose a novel dynamic evolution model called clock finite state automata (CFSA) using formal methods, exhibiting the overall changing processes of global clock states. We also develop a clock consistency evolution algorithm using CFSA to quantitatively evaluate the influencing factors. The experimental evaluation shows that network delay (41.4% on average) is the greatest impact factor; the frequent entry and exit of the sensor nodes (29.9%) are the second greatest, and the oscillator jitter of computers (11.1%) is the least impact factor. Compared with traditional analysis methods, our method has good feasibility, effectiveness, and novelty. The analysis results can guide the designers of new clock synchronization algorithms for distributed sensor nodes in IoT.