發(fā)布時間：2018-05-04 20:18

  本文選題：分布式計算 + 無線網(wǎng)絡(luò)　； 參考：《南京大學》2014年碩士論文

【摘要】：隨著無線網(wǎng)絡(luò)和移動計算技術(shù)的迅速發(fā)展,整個網(wǎng)絡(luò)的狀態(tài)呈現(xiàn)出兩大趨勢。首先,網(wǎng)絡(luò)中的用戶數(shù)量迅速增加。當使用信道的用戶越來越多,而信道頻譜作為一種不變的自然資源顯得越來越稀缺的時候,一種高效的調(diào)度算法顯得越來越重要。其次,用戶的行為逐漸由硬件控制轉(zhuǎn)向軟件控制,從而變得越來越智能。用戶可以根據(jù)自己的需求破壞既定的通信協(xié)議,做出更加復(fù)雜的行為,從而增加自己所獲得的網(wǎng)絡(luò)資源。這樣的行為有可能損害其他用戶的利益,并且破壞整個系統(tǒng)的性能。本篇論文將該優(yōu)化問題建模成為圖博弈,目標是使用局部的分布式算法給出一個性能最優(yōu)或者接近最優(yōu)的納什均衡(Nash equilibrium)。納什均衡的最優(yōu)性保證了整個網(wǎng)絡(luò)的性能,納什均衡的穩(wěn)定性保證了用戶不會偏離已有的協(xié)議,從而同時克服了上述兩個問題。為了實現(xiàn)這個目標,我們將圖博弈的優(yōu)化問題歸約為在相應(yīng)的因子圖中計算最大后驗概率,并且使用置信傳播算法(Belief Propagation)來設(shè)計局部的消息傳遞算法。該算法在無環(huán)圖上可以在線性時間內(nèi)求出最優(yōu)解,在有環(huán)圖上同樣可以保證極快的收斂速度和穩(wěn)定的性能。此外,在平面圖上,該算法可以給出極好的近似結(jié)果。
[Abstract]:With the rapid development of wireless network and mobile computing technology, the state of the whole network presents two major trends. First, the number of users in the network is growing rapidly. As more and more users use the channel, and the channel spectrum is becoming more and more scarce as a kind of invariant natural resources, an efficient scheduling algorithm becomes more and more important. Secondly, the behavior of users is gradually changed from hardware control to software control, thus becoming more and more intelligent. Users can break the established communication protocols and make more complex behaviors according to their own requirements, thus increasing the network resources they obtain. Such behavior may harm the interests of other users and undermine the performance of the system as a whole. In this paper, the optimization problem is modeled as a graph game, and the goal is to give a Nash equilibrium Nash equilibrium with or near optimal performance using a local distributed algorithm. The optimality of Nash equilibrium guarantees the performance of the whole network, and the stability of Nash equilibrium ensures that users will not deviate from the existing protocols, thus overcoming the above two problems simultaneously. In order to achieve this goal, we reduce the optimization problem of graph game to calculate the maximum posterior probability in the corresponding factor graph, and design a local message passing algorithm by using the confidence propagation algorithm (belief Propagation). The algorithm can obtain the optimal solution in linear time on the acyclic graph, and it can also guarantee the extremely fast convergence speed and stable performance on the annular graph. In addition, the algorithm can give excellent approximate results on planar graphs.
【學位授予單位】：南京大學
【學位級別】：碩士
【學位授予年份】：2014
【分類號】：TN911.5

【相似文獻】

相關(guān)期刊論文 前2條

1 龍航;曠婧華;申山山;鄭侃;王文博;;迫零中繼系統(tǒng)中的空間信道配對與映射[J];北京郵電大學學報;2010年05期

2 ;[J];;年期

相關(guān)會議論文 前1條

1 李萍;梁昌洪;張殿富;;利用“極化差異”構(gòu)造獨立空間信道的試驗與研究[A];2003'全國微波毫米波會議論文集[C];2003年

相關(guān)重要報紙文章 前1條

1 ;美國愛瑞通信將先進的無線技術(shù)帶入生活[N];中國高新技術(shù)產(chǎn)業(yè)導(dǎo)報;2001年

相關(guān)碩士學位論文 前1條

1 王逸愷;空間信道博弈的分布式算法[D];南京大學;2014年

，

本文編號：1844496