Adaptive Control of Nonstatistical Sensor Data Aggregation to Minimize Latency in IoT Gateways

Abstract

In order to satisfy rigorous quality of service requirements and realize latency-critical Internet of Things (IoT) applications, it is important to minimize the latency at IoT gateways, which aggregate a large amount of small-sized data from massive sensor devices. Data aggregation in IoT gateways can be broadly classified into statistical and nonstatistical. In previous studies on statistical data aggregation, we derived the Laplace-Stieltjes transform of the latency distributions and accurate estimation formulae for the optimal aggregation parameters under steady-state conditions with a Poisson arrival. Moreover, we proposed an adaptive control scheme of statistical data aggregation that minimizes the latency when time variation exists in the arrival rate. Furthermore, we analyzed the nonstatistical aggregation and derived an approximation of the average latency. In this study, we apply the approximation to the adaptive control for the nonstatistical aggregation based on the time-variant arrival rate and propose three kinds of estimation formulae for the optimal aggregation number. The transient and average characteristics of the estimation formulae were compared by simulation. The results indicated that the proposed control with each estimation formula achieved stable and nearly theoretically optimal latency.