【摘要】：基于譜域變換的圖上信號處理是網(wǎng)絡(luò)化數(shù)據(jù)處理中的新興理論.許多圖上信號處理應(yīng)用中假設(shè)圖上信號的頻譜以低頻成分為主,而實際上圖的邊權(quán)重對于圖上信號的頻譜分布有重要影響,因此本文研究圖的構(gòu)造過程中的邊權(quán)重設(shè)計問題.首先將其轉(zhuǎn)化為最優(yōu)化問題,并得到了最優(yōu)邊權(quán)重的解析式.進一步針對網(wǎng)絡(luò)化數(shù)據(jù)處理中分布式計算的需求,提出了基于網(wǎng)絡(luò)數(shù)據(jù)的分布式權(quán)重優(yōu)化算法.仿真結(jié)果表明相比于傳統(tǒng)的邊權(quán)重構(gòu)造方法,本文提出的算法能夠使信號的能量更多地集中在圖譜的低頻部分.
[Abstract]:Map signal processing based on spectral domain transformation is a new theory in networked data processing. In many signal processing applications, it is assumed that the frequency spectrum of the signal on the map is dominated by the low frequency component, but in fact the edge weight of the graph has an important effect on the spectrum distribution of the signal on the map. Therefore, the design of the edge weight in the construction of the graph is studied in this paper. Firstly, it is transformed into an optimization problem, and the analytical formula of the optimal edge weight is obtained. To meet the demand of distributed computing in networked data processing, a distributed weight optimization algorithm based on network data is proposed. The simulation results show that the proposed algorithm can concentrate the energy of the signal in the low-frequency part of the map more than the traditional method of edge weight construction.
【作者單位】： 復(fù)旦大學(xué)電子工程系;復(fù)旦大學(xué)電磁波信息科學(xué)教育部重點實驗室;
【基金】：國家科技重大專項(2015ZX03001040)
【分類號】：TN911.7
，

本文編號：2327630