Search for: All records

Information fusion is a procedure that merges information locally contained at the nodes of a network. Of high interest in the field of distributed estimation is the fusion of local probability distributions via a weighted geometrical average criterion. In numerous practical settings, the local distributions are only known through particle approximations, i.e., sets of samples with associated weights, such as obtained via importance sampling (IS) methods. Thus, prohibiting any closed-form solution to the aforementioned fusion problem. This article proposes a family of IS methods—called particle geometric–average fusion (PGAF)—that lead to consistent estimators for the geometrically-averaged density. The advantages of the proposed methods are threefold. First, the methods are agnostic of the mechanisms used to generate the local particle sets and, therefore, allow for the fusion of heterogeneous nodes. Second, consistency of estimators is guaranteed under generic conditions when the agents use IS-generated particles. Third, a low-communication overhead and agent privacy are achieved since local observations are not shared with the fusion center. Even more remarkably, for a sub-family of the proposed PGAF methods, the fusion center does not require the knowledge of the local priors used by the nodes. Implementation guidelines for the proposed methods are provided and theoretical results are numerically verified. 

In this work, we consider a distributed online convex optimization problem, with time-varying (potentially adversarial) constraints. A set of nodes, jointly aim to minimize a global objective function, which is the sum of local convex functions. The objective and constraint functions are revealed locally to the nodes, at each time, after taking an action. Naturally, the constraints cannot be instantaneously satisfied. Therefore, we reformulate the problem to satisfy these constraints in the long term. To this end, we propose a distributed primal-dual mirror descent-based algorithm, in which the primal and dual updates are carried out locally at all the nodes. This is followed by sharing and mixing of the primal variables by the local nodes via communication with the immediate neighbors. To quantify the performance of the proposed algorithm, we utilize the challenging, but more realistic metrics of dynamic regret and fit. Dynamic regret measures the cumulative loss incurred by the algorithm compared to the best dynamic strategy, while fit measures the long term cumulative constraint violations. Without assuming the restrictive Slater’s conditions, we show that the proposed algorithm achieves sublinear regret and fit under mild, commonly used assumptions.