The Value of Information in Distributed Decision Networks
Abstract
We now define a mathematical framework for networks. Let G = (V,E) be an undirected random network (graph) drawn from a known distribution pG,1 composed of a finite vertex set V and a link set E = { V x V modulo S} where S = {{(i; j); (j; i)}}. Each vertex i isin V corresponds to an agent, and each link (j; i) isin E corresponds to a channel by which information flows from agent j to agent i in the network. We denote the neighborhood of I by N(i) = {j vertical line (i; j) isin E}.There is a state (internal or external) W drawn from a distribution pW that the agents may want to estimate, transmit, and act upon. Each agent i also possesses a state and some private observation about W. We denote the state at time t by xi(t), and we assume that the tuple of initial states (xi(0)) is correlated with W and are drawn randomly from a joint distribution pWX0 . Agent i's private information/observation at time t is denoted by Yi(t) and has a joint distribution pWYi(t) with W. Finally, agent i has some information about agent j's state (either because it can observe it or agent j transmits it), which we denote by mji(t) = mji(xj(t)).
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 04, 2016
- Accession Number
- AD1031881
Entities
People
- Dahleh Munther
Organizations
- Massachusetts Institute of Technology