考慮服務品質與訊框分配之OFDMA排程演算法

正交分頻多工存取（OFDMA）已經成為一個在無線通訊系統上非常受歡迎的技術，如IEEE 802.16e標準而OFDMA的框架結構在近年來則引發了一系列的研究趨勢。在IEEE 802.16e標準，每一次我們傳送給一個特定用戶的突發性資料訊框必須是由子通道和時間槽所形成的二維矩形訊框。如果太多的使用者資料積壓在同一時間，這樣一來並非全部的使用者資料都會在下一個訊框中被傳送出去，反而還會造成而外的延遲發生。換句話說，該系統應有效的在OFDMA技術下對這些使用者資料作訊框排程，以確保延遲的上限和維持較高的吞吐量。儘管先前的研究已經在解決在OFDMA突發性資料訊框的封包排程，而我們提出了新的下行訊框分配演算法讓其在使用基因演算法及貨板裝載演算法的情況下決定在OFDMA技術下可傳輸的最大訊息量。因外，我們考慮了在正交分頻多工存取無線系統下不同應用的服務品質所需。為了這個目的，我們將結合服務品質和分配演算法並且將其變得更為可行。而我們的實驗結果將會證明，當我們使用交配次數大於5次的基因演算法，我們可以充分使用至少86%位於正交分頻多工存取無線系統訊框中的時槽。此外，如果我們遇到過多的封包湧入緩衝區時，我們可以減少基因演算法的交配次數已利於減少系統的運算量。另外，我們採取隨機網路模型估測封包的延誤時間。對照於決定型網路模型，我們的隨機網路摸型可以更精準地估測到封包的延誤時間。

關鍵字

服務品質；正交分頻多工存取；排程

並列摘要

Orthogonal Frequency Division Multiple Access (OFDMA) has become a very popular technique in wireless communication systems such as the IEEE 802.16e and the OFDMA frame structure has triggered a series of research trends in recent years. In the IEEE 802.16e standard, every burst we transmitted to a particular user must be a rectangle in a two-dimensional frame constructed by sub-channel and time slot. If too many bursts are backlogged at the same time, not all bursts can be sent in the next frame and additional delay occurs. In other words, the system should arrange the burst transmitted in an OFDMA frame efficiently in order to assure limit delay and maintain high throughput. Although there are previous works in resolving the packet scheduling of OFDMA bursts, we proposed a new downlink map allocated algorithm to decide the maximum information bits which can be transmitted in one OFDMA frame by using the technique of genetic algorithm and heuristic pallet loading algorithm. Furthermore, we consider various QoS requirements for different applications in OFDMA wireless communication systems. For this purpose, we combine QoS and allocation algorithm to make the proposed design more practical. The simulation results show that we can allocate at least 86% of slots in an OFDMA frame when the number of genetic generations is more than 5. Moreover, we can execute less times in genetic algorithm to reduce the computational time when there are too many packets in the buffer. In addition, we successfully adopt the stochastic-network-server model to estimate the resulting packet delay. Comparing with deterministic-network-server model, our stochastic-network-server model can provide predictions more explicitly about its probabilistic behavior of packet delay in any threshold.

並列關鍵字

QoS ； OFDMA ； Scheduling ； 802.16e

參考文獻

[1] L. Wan, W. Ma, Z. Guo. “A Cross-layer Packet Scheduling and Subchannel Allocation Scheme in 802.16e OFDMA System, ” IEEE WCNC 2007, pp.1867－1872.

[2] R. Cohen L. Katzir. “Computational analysis and efficient algorithms. for micro and macro ofdma scheduling,” IEEE INFOCOM, pp. 511-519, 2008 .

[3] Y. Cui, T. Gu, W. Hu. « An Algorithm for the Constrained Two-Dimensional Rectangular Multiple Identical Large Object Placement Problem,” Optimization Methods and Software. 2008.

[4] Y. Cui, T. Gu, W. Hu. Simplest optimal guillotine cutting patterns for strips of identical circles. Journal of Combinatorial Optimization. 2008.

[5] Y. Cui. “An exact algorithm for generating homogeneous two-segment cutting patterns,” Engineering Optimization. 2007.

國際替代計量

考慮服務品質與訊框分配之OFDMA排程演算法

全文下載

主題瀏覽