| 摘要: | 在本篇論文中，為了要解決叢集小型基地台間的互相干擾，提出了將拍賣理論應用於系統資源分配的想法；有別於傳統的ICIC/eICIC演算法雖然觀念簡單易於實做，但對頻譜的使用不夠有彈性，因此，希望利用拍賣"競爭"的概念讓最合適的人來使用資源，讓資源利用最佳化，在此先以第一密封式拍賣為基礎提出了投標式拍賣演算法，其中以電信營運商為賣方，系統資源(Resource Blocks)為拍賣品，各個小型基地台(Smallcells)為投標者，並選擇讓叢集中的一個小型基地台當任拍賣仲裁者角色來接受其他基地台的出價以及裁定投標勝負；各個基地台的出價策略是根據使用系統資源的效率為標準，讓可得到更高throughput的基地台會有較大的優勢去贏得競價，希望運用"適材適用"的觀念，進而提升整體系統效能；；除此之外，考慮到投標式拍賣演算法的運用將會集中大量運算於單一節點上，可能會造成擔任拍賣官的基地台運算量過大、無線資源不足、公平性問題、甚至因為天災人禍造成仲裁者損壞使系統不正常運作，為了改善這些問題，將原本演算法改變拍賣的策略與方式，變成以英式拍賣為基礎，利用所有會互相干擾的基地台皆是受同一電信商控制的概念，制訂所有人都必須尊崇的出價原則，藉此取消掉拍賣仲裁者角色，並在投標者的出價原則中加入了平衡持有比例的想法，改善投標式拍賣演算法中的投標者滿意度問題，分散式拍賣演算法的最終目標是"在追求高系統效能的情況下，保留有一定的資源給予弱勢的拍賣參與者"。;For the next generation network technology - Internet of Thing(IOT), which means everything in our daily life will connect to internet, for example: television, air conditioner, refrigerator, medicine equipment, and so on. Those of them hunger for network serving will cause an evolution to wireless network communication technology.
Adding more small cells, which are lower power, shorter coverage than the original Marco-cells in heterogeneous network architecture may be one of solutions to fulfill those requirements. However, numerous small cells will form a cluster, and they may cause additional problem to each member, i.e. the interference. The inter-cell interference coordinated(ICIC) was used to eliminate interference among Marco-cells in the 3GPP specifications of Release 8 about LTE can be adapted to release the interference problem. It is easy to understand and achieve, but its system resources utilization isn′t as same as we want. We dedicate ourselves to search for another possibility.
In this dissertation, we try to apply auction theory′s concept into addressing the interference among a cluster of small cells. We will propose an algorithm based on sealed-bid first price auction. In this auction algorithm, those of small cells can fight for system resource based on their channel quality. We hope the outcome is that system resource will be used by the best-fit base station, then that is why system throughput in auction algorithm is much higher than ICIC resolution.
Although the former auction algorithm gets high performance, we find that there are some small cell eNodeBs can′t grasp any resource during the auction in some conditions. The unfair defeat comes from the similar wanting resource and the small cell base station is not outstanding. These problem is the opportunity to make us to change our idea. We dramatic modify our algorithm, the distributed auction algorithm, which is based on English auction and erase the auction judge in new version. Concept of holding fairness is introduced in small cells′ bid process, it means that if one of small cell in the cluster doesn′t get enough resource in this auction, it will be assigned more power to compete for resource in the next auction. Distributed auction algorithm not only keeps advantages of the high system throughput compared to that of the former auction algorithm, but also delete the unfair issue. |
