- 學位論文
具能量效率的多重存取協定設計
Energy efficiency multiple access protocol design
摘要
在這篇論文中，我們提出一個在無限資料流的假設下適用於各種不同的多重存取協定的整合數學模型，並且以設計能量效率最佳化為我們的目的。我們利用嵌入馬爾科夫鏈(embedded Markov chain)廣泛地模擬ALOHA、Non-persistent CSMA、1-persistent CSMA和GRAP協定來得到系統總處理能力、延遲時間和能量效率等性能。然而，無限資料流的假設隱含著表示退避時間和總計抵達速率無關。我們放鬆待傳使用者的無關性並且利用提出的整合MAC演算法的原型來引導出協定和退避機制的關聯性。以直覺來說，增加平均退避時間也會跟著增加延遲時間。但在同時，它會降低總計抵達速來降低重傳次數。這個利益交換讓我們把調整退避時間視為最佳化能量效率的方法。因此，我們提出能動態調整延遲時間並以整合樹狀協定為基礎的最佳樹狀協定。我們也發現能量效率和延遲時間擁有相同的表達式-我們稱做加權時間。他們只差別在參數設定不同而已。然而，根據不同的系統限制假設，最佳化準則也將不同。我們將顯示出在一般限制下能量效率最佳化的數值分析結果。
並列摘要
In this thesis, we propose a general mathematic model under infinite traffic load for various multiple access protocols and aim at the designing of optimization of energy efficiency. We generally model the ALOHA, Non-persistent and 1-persistent CSMA, and GRAP protocols using embedded Markov chain to obtain the performances of throughput, delay and energy efficiency. However, the infinite traffic loads implicit express the independence of backoff time and aggregate arrival rate. We loosen this independence of backlogged users and propose the prototype of unified MAC protocol which shows the correlation of protocol and backoff mechanism. In intuition, increase average backoff time increases the delay as well. In the mean while, it decrease the aggregate arrival rate which result in less retransmission times. This tradeoff leads us to consider the optimal energy efficiency by adjusting backoff time. Thus, we propose optimal tree protocol based on the unified tree protocol by adjusting dynamic backoff time. We also figure out that energy efficiency and delay are based on the general form which we called weighted time. They are only different in parameters setting. Nevertheless, according to unlike assumptions of system constrain, the optimization criterion is also distinct. We show the optimal energy efficiency under general constrain by numerical result.