長期演進系統中機器對機器通訊之壅塞控制、情境估計，與基於賽局理論的資源配置

Abstract

行動裝置、機器對機器通訊、物聯網裝置在數量上以及產生的流量上的爆炸性成長，有很大可能性產生數量龐大的網路連線請求，因此引起了人們對於無線電接取網路的容量的注意。大量湧現的隨機存取需求不止導致嚴重的隨機存取前導信令碰撞，也造成下載資源匱乏，因此降低了隨機存取程序的效能。然而，下載資源匱乏對於系統效能的影響目前尚未被完整的研究。此外，大部分現有既存的隨機存取雍塞解決方案是用犧牲隨機存取率來換取較高的隨機存取成功機率，其代價就是低隨機存取率的方法需要很長的時間來處理所有接取的需求。 本博士論文，目的在解決機器對機器通訊的情境下所發生的隨機存取過載問題，由三個階段組成。在第一階段，我們總結了最好的隨機存取壅塞控制方法，並且藉由模擬深入研究它們的效能。經由第一階段的的研究，我們認為隨機存取的壅塞源自於過高的隨機存取前導信令傳輸率，也就是太多裝置同時競爭隨機存取資源。此外，大部分現存的方法都在隨機存取的成功機率與資源使用率這兩方面有所妥協。在第二階段，我們提出了一個隨機存取需求的估計方法，來估計有多少使用者裝置試圖進行網路連線。 有了這個估計的資訊，基地台就可以在保證滿意的隨機存取成功機率的情況下，大幅提升隨機存取率。此外，我們研究了下載資源匱乏與前導信令偵測機率對於隨機存取效能的影響。我們也得到了隨機存取延遲、隨機存取資源量、與隨機存取資源效率這三者的關係。在第三階段，我們使用賽局理論來處理不同種類裝置間的資源分配問題。我們研究機器對機器通訊與人類通訊在隨機存取方面的共存問題。我們也提出一個拍賣模型來處理不同機器對機器通訊類別的隨機存取流量在不同的時間區間上的分配問題，並且藉由第二階段提出的隨機存取估計方法，讓本拍賣模型所建議的最佳流量分配可以被實現。

這三階段的研究為隨機存取的壅塞控制提供了一套完整的解決方案，整合負載估計與資源分配的方法。所提出的負載估計方法可以在極少信另花費的情況下優異的運作，而資源分配的方法更可以在保證隨機存取成功機率的情況下，大幅提升隨機存取的資源效率。此外，我們也對集中式與分散式的資源分配架構提出新穎的設計，包括集中式的拍賣機制，與分散式的非合作賽局模型。 本博士論文的研究成果，詳實的提出關鍵性的負載估測技術與實用的資源配置方法，預期能在未來的蜂巢式物聯網網路中，對突然湧現且巨量的機器對機器隨機存取需求提供強而有力的支援。

With the potential to generate numerous connection requests, an explosive growth in the volume of data traffic and the number of mobile, machine-to-machine (M2M), and Internet-of-Things (IoT) devices has drawn new attention on the capacity of radio access network (RAN). Surging random access attempts cause not only severe preamble collisions but also downlink resource shortage, and thus degrade the performance of random access procedure. However, the effect of downlink resource shortage on system performance is not yet comprehensively studied. In addition, most existing random access contention resolution mechanisms sacrifice RACH (random access channel) throughput for a high success probability, and thus the price is that low-throughput mechanisms need long time to deal with access attempts. This dissertation, aiming to resolve RACH overload problems in M2M context, consists of 3-phase research. In the phase I, we summarize the the-state-of-the-art methods for RACH overload control, and investigate their performance through simulations. Through Phase I research, we conclude that RACH contention stems from a high preamble transmission rate, i.e., too many devices send their preambles in the same random access resources. In addition, most existing methods have a compromising tradeoff between RACH success probability and RACH resource utilization. In phase II, we proposes a RACH attempt estimation method to estimate the number of UEs that are trying for network connection. With the estimated information, the base station can boost the RACH throughput while guaranteeing satisfactory RACH success probability. Moreover, we obtain the effect of downlink resource shortage and preamble detection probability on RACH performance. We also find the relation among access delay, RACH throughput, and resource efficiency. In phase III, we apply game theory to address the resource allocation problem among different types of devices. We study the coexistence problem of (H2H) human-to-human and M2M in RACH contention. We also propose an auction model to allocate RACH traffic from multiple M2M applications on multiple transmission periods, and make it implementable through the application of RACH attempt estimation proposed in phase II. The 3-phase study provides a complete suit of solutions for RACH congestion control integrating RACH load estimation and RACH resource allocation. The proposed load estimation method works quite efficiently with little signaling cost, while the designed resource allocation methods significantly boost RACH throughput while guaranteeing RACH success probability. Moreover, we propose novel designs for both centralized and distributed resource allocation framework, i.e., centralized auction scheme, and distributed non-cooperative game model. The research fruit of the dissertation, including key load estimation technique and practical resource allocation modelling, is expected to powerfully support the management of bursty and massive M2M random accesses in the future cellular IoT networks.