- 學位論文
粒子群演算法與模糊邏輯控制基於負載平衡及資源分配應用於無線通訊網路
Applications of PSO and FLC based on load balancing and resource allocation in wireless communication networks
摘要
近年來，伴隨科技技術的發展，使用者對於行動設備及網路的依賴程度明顯提高。在行動設備使用時的電池續航時間中，行動設備產品的電池容量發展受到產品大小限縮，所以如何減少設備使用行動網路時的電力消耗及提升網路資源的分配成為一個重要的研究議題。在這些改善方法中，其中Dual-Radio Opportunistic Networking for Energy Efficiency - Weighted Round Robin (DRONEE-W)為一種混合型的網路架構，這個架構將行動設備利用無線感測網路(Wireless Sensor Networks, WSNs)的叢集化節點的概念，劃分行動設備成多個叢集，以及採用機會型選擇方法，挑選行動設備與LTE-A基地台(Base Station, BS)之間較高通訊號品質的行動設備節點成為叢集頭，透過關閉非叢集頭的LTE-A功能，叢集內部通訊採用Wi-Fi傳遞封包，減少行動設備與LTE-A基地台的通訊時的電量消耗，同時也可降低行動裝置彼此間訊號的干擾。在本篇論文我們提出的DRONEE2-W、DRONEE-PF2-PW及DRONEE-PFM-PWRC三個方法，DRONEE2-W為DRONEE-W的延伸，方法採用2-Hop叢集架構改善原先的電量消耗。第二方法DRONEE-PF2-PW為DRONEE2-W的延伸，在選擇叢集頭方面，我們會利用粒子群演算法(Particle Swarm Optimization, PSO) 尋找叢集頭的最佳位置，並透過模糊邏輯控制(Fuzzy Logic Control, FLC)調配PSO的Cost Function參數值，及延續DRONEE2-W中2-Hop的叢集架構，在增加叢集的覆蓋率同時也提高行動設備的負載平衡，在LTE-A基地台與叢集的網路資源分配中，引入服務品質(Quality of Service, QoS)將封包依據資料型態分類優先權權重，提出優先權加權輪調法(Priority Weighted Round Robin, PWRR)改善參考方法的加權輪調法(Weighted Round Robin, WRR)，提高網路資源分配的Uplink吞吐量。第三方法DRONEE-PFM-PWRC為DRONEE-PF2-PW的改善，透過引入MHDEECFLC方法中允許偏遠節點擴大廣播範圍加入叢集的概念來，來平衡偏遠節點電量消耗過快的問題，在LTE-A基地台與叢集之間的網路資源分配中，提出優先權加權速率調整方法(Priority Weighted Rate Control, PWRC)提升網路資源分配。我們的三個方法有效的減少6%至14%的電量消耗及提升3%至13%的網路吞吐量及37%至79%的傳送延遲時間。
並列摘要
As technology develops, people become relying more and more on smartphones and internet in recent years. For this purpose, extending battery life is a crucial problem for current smartphones. Although the battery capacity on smartphones are making progress, but the capacity levels of the smartphones are limited due to the dimension of the battery. Therefore, how to balance the internet efficiency and increase the internet throughput, also reduce the consumption of battery energy are the fundamental issue. On the ways to solve this problem, one of them is the Dual-Radio Opportunistic Networking for Energy Efficiency – Weighted Round Robin (DRONEE-W). DRONEE-W is a type of hybrid network that introducing the Cluster concept of Wireless Sensor Networks (WSNs) divides the smartphones into several clusters, and it selects the smartphones with better telecommunication quality to be the cluster head. The smartphone within the cluster will shut off the Long Term Evolution - Advanced (LTE-A) function to reduce the energy consumption in telecommunications with the BS, and share the internet source with the cluster head through Wi-Fi in the cluster group. It can reduce the interference signals affecting from other smartphones and reduce the energy consumption of the mobile device as well. In this thesis, we propose three schemes DRONEE2-W, DRONEE-PF2-PW and DRONEE-PFM-PWRC. DRONEE2-W is the extension of reference method DRONEE-W. Then, we add 2-Hop concept of Multi-Hop into the cluster group to increase the coverage for the cluster. The second method DRONEE-PF2-PW is the extension of first method DRONEE2-W. In choosing the cluster head, we use Particle Swarm Optimization (PSO) to combine Fuzzy Logic Control (FLC) with Cost Function parameters in PSO so as to look for the best cluster head and decrease the problem of less membership. We extend 2-Hop cluster architecture from DRONEE2-W. In the aspect of resource distribution among cluster and BS, we combine Quality of Service (QoS) to classify flow categories into priority-weighted, and then using the priority-weighted outcome to improve the allocation method Weighted Round Robin (WRR) as Priority Weighted Round Robin (PWRR). The PWRR can enhance uplink throughput in resource allocation. The DRONEE-PFM-PWRC is our DRONEE-PF2-PW upgrade. We introduce the concept of MHDEECFLC that allowing remote node extend its broadcast range to join adjacent cluster to balance energy consumption. In the aspect of resource distribution among cluster and BS, we propose Priority Weighted Rate Control (PWRC) to improve resource allocation. Our proposed three methods can reduce network consumption of 6% to 14%, enhance network throughput of 3% to 13% and reduce transmission delay of 37% to 79%.