Optimal Graph-Filter Design and Applications to Distributed Linear Network Operators

被引:194
作者
Segarra, Santiago [1 ]
Marques, Antonio G. [2 ]
Ribeiro, Alejandro [3 ]
机构
[1] MIT, Inst Data Syst & Soc, 77 Massachusetts Ave, Cambridge, MA 02139 USA
[2] King Juan Carlos Univ, Dept Signal Theory & Commun, Madrid 28943, Spain
[3] Univ Penn, Dept Elect & Syst Engn, Philadelphia, PA 19104 USA
关键词
Graph signal processing; graph filter (GF) design; distributed linear transformation; network operator; finite-time consensus; analog network coding; CONSENSUS; ALGORITHMS;
D O I
10.1109/TSP.2017.2703660
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
We study the optimal design of graph filters (GFs) to implement arbitrary linear transformations between graph signals. GFs can be represented by matrix polynomials of the graph-shift operator (GSO). Since this operator captures the local structure of the graph, GFs naturally give rise to distributed linear network operators. In most setups, the GSO is given so that GF design consists fundamentally in choosing the (filter) coefficients of the matrix polynomial to resemble desired linear transformations. We determine spectral conditions under which a specific linear transformation can be implemented perfectly using GFs. For the cases where perfect implementation is infeasible, we address the optimization of the filter coefficients to approximate the desired transformation. Additionally, for settings where the GSO itself can be modified, we study its optimal design as well. After this, we introduce the notion of a node-variant GF, which allows the simultaneous implementation of multiple (regular) GFs in different nodes of the graph. This additional flexibility enables the design of more general operators without undermining the locality in implementation. Perfect and approximate designs are also studied for this new type of GFs. To showcase the relevance of the results in the context of distributed linear network operators, this paper closes with the application of our framework to two particular distributed problems: finite-time consensus and analog network coding.
引用
收藏
页码:4117 / 4131
页数:15
相关论文
共 39 条
  • [1] Network information flow
    Ahlswede, R
    Cai, N
    Li, SYR
    Yeung, RW
    [J]. IEEE TRANSACTIONS ON INFORMATION THEORY, 2000, 46 (04) : 1204 - 1216
  • [2] [Anonymous], 2001, ALGEBRAIC GRAPH THEO
  • [3] DISTRIBUTED SIGNAL SUBSPACE PROJECTION ALGORITHMS WITH MAXIMUM CONVERGENCE RATE FOR SENSOR NETWORKS WITH TOPOLOGICAL CONSTRAINTS
    Barbarossa, S.
    Scutari, G.
    Battisti, T.
    [J]. 2009 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH, AND SIGNAL PROCESSING, VOLS 1- 8, PROCEEDINGS, 2009, : 2893 - 2896
  • [4] Bertsekas D. P., 1989, PARALLEL DISTRIBUTED, V23
  • [5] METHOD OF CENTERS FOR MINIMIZING GENERALIZED EIGENVALUES
    BOYD, S
    ELGHAOUI, L
    [J]. LINEAR ALGEBRA AND ITS APPLICATIONS, 1993, 188 : 63 - 111
  • [6] Chen SH, 2015, INT CONF ACOUST SPEE, P3731, DOI 10.1109/ICASSP.2015.7178668
  • [7] Discrete Signal Processing on Graphs: Sampling Theory
    Chen, Siheng
    Varma, Rohan
    Sandryhaila, Aliaksei
    Kovacevic, Jelena
    [J]. IEEE TRANSACTIONS ON SIGNAL PROCESSING, 2015, 63 (24) : 6510 - 6523
  • [8] Golub GH., 2012, MATRIX COMPUTATIONS, V3
  • [9] Ho T., 2008, Network Coding: An Introduction
  • [10] Isufi E, 2016, EUR SIGNAL PR CONF, P200, DOI 10.1109/EUSIPCO.2016.7760238