利用時空訊號統計相關性之物聯網預測式媒體接取機制

Abstract

本篇論文當中，我們提出了運用在物聯網之下，利用時空訊號統計相關性預測式媒體接取機制。由於大量的通訊設備(MTC設備)及有限的媒介存取資源，基地台媒介存取資源無法有效地分配在各個MTC設備。基於每個MTC設備對於延遲的容忍性不同以及MTC設備都以上傳資料為主，我們希望從空間的相關特性以及訊號的觀點，從眾多的MTC設備中，找出擁有重要資訊的MTC設備，並將有限的媒介存取資源分配給擁有重要資訊的這些設備。若一個MTC設備的訊號與他之前的訊號有很大的差異，其訊號對基地台視為比較重要的。我們藉由隨機程序中的自我迴歸模型建立每個MTC設備的訊號，以及建立MTC設備之間的空間相關特性，基地台可以根據之前收到的訊號以及MTC設備之間的空間相關特性來預測一個MTC設備現在的訊號，並且將媒體接取機制的資源優先分配給預測誤差較大的MTC設備，如此便可降低系統的總體預測誤差。經由數學推導降低運算的複雜度，我們以一階的自我迴歸模型加上時空的相關特性來實現我們的預測式輪詢演算法，並模擬結果來證明所提出的演算法相較於輪詢演算法是較好的

In this thesis, we propose spatial and temporal correlations for signal-centric predictive polling for medium access control in the machine-to-machine communication networks. Due to the lots of MTC devices and the limited resource for medium access control, it is not feasible for all machines to send data successfully in one time slot. Many previous works on medium access control do not take into consideration of the values of the signal. The proposed signal-centric predictive polling schemes always collect the most valuable signal. We construct the model by using autoregressive model to calculate the mean squared error (MSE) of prediction, and exploit the statistical correlations of regular processes and the characteristics of the spatial relation on M2M communication network. We adopt the widely used autoregressive (AR) model to calculate the mean squared error (MSE) of prediction based on the polling decision. We formulate a discrete optimization problem to make the optimal polling decision in one time slot, and justify the proposed algorithm by simulation to show our algorithm could significantly outperform the round-robin scheme and temporal correlation for signal-centric predictive polling algorithm.