In this paper, a general nonlinear 1st-order consensus-based solution for distributed constrained convex optimization is considered for applications in network resource allocation. The proposed continuous-time solution is used to optimize continuousl
y-differentiable strictly convex cost functions over weakly-connected undirected multi-agent networks. The solution is anytime feasible and models various nonlinearities to account for imperfections and constraints on the (physical model of) agents in terms of their limited actuation capabilities, e.g., quantization and saturation constraints among others. Moreover, different applications impose specific nonlinearities to the model, e.g., convergence in fixed/finite-time, robustness to uncertainties, and noise-tolerant dynamics. Our proposed distributed resource allocation protocol generalizes such nonlinear models. Putting convex set analysis together with the Lyapunov theorem, we provide a general technique to prove convergence (i) regardless of the particular type of nonlinearity (ii) with weak network-connectivity requirement (i.e., uniform-connectivity). We simulate the performance of the protocol in continuous-time coordination of generators, known as the economic dispatch problem (EDP).
Observability and estimation are closely tied to the system structure, which can be visualized as a system graph--a graph that captures the inter-dependencies within the state variables. For example, in social system graphs such inter-dependencies re
present the social interactions of different individuals. It was recently shown that contractions, a key concept from graph theory, in the system graph are critical to system observability, as (at least) one state measurement in every contraction is necessary for observability. Thus, the size and number of contractions are critical in recovering for loss of observability. In this paper, the correlation between the average-size/number of contractions and the global clustering coefficient (GCC) of the system graph is studied. Our empirical results show that estimating systems with high GCC requires fewer measurements, and in case of measurement failure, there are fewer possible options to find substitute measurement that recovers the systems observability. This is significant as by tuning the GCC, we can improve the observability properties of large-scale engineered networks, such as social networks and smart grid.
This paper considers distributed estimation of linear systems when the state observations are corrupted with Gaussian noise of unbounded support and under possible random adversarial attacks. We consider sensors equipped with single time-scale estima
tors and local chi-square ($chi^2$) detectors to simultaneously opserve the states, share information, fuse the noise/attack-corrupted data locally, and detect possible anomalies in their own observations. While this scheme is applicable to a wide variety of systems associated with full-rank (invertible) matrices, we discuss it within the context of distributed inference in social networks. The proposed technique outperforms existing results in the sense that: (i) we consider Gaussian noise with no simplifying upper-bound assumption on the support; (ii) all existing $chi^2$-based techniques are centralized while our proposed technique is distributed, where the sensors textit{locally} detect attacks, with no central coordinator, using specific probabilistic thresholds; and (iii) no local-observability assumption at a sensor is made, which makes our method feasible for large-scale social networks. Moreover, we consider a Linear Matrix Inequalities (LMI) approach to design block-diagonal gain (estimator) matrices under appropriate constraints for isolating the attacks.
In this paper, we consider the binary classification problem via distributed Support-Vector-Machines (SVM), where the idea is to train a network of agents, with limited share of data, to cooperatively learn the SVM classifier for the global database.
Agents only share processed information regarding the classifier parameters and the gradient of the local loss functions instead of their raw data. In contrast to the existing work, we propose a continuous-time algorithm that incorporates network topology changes in discrete jumps. This hybrid nature allows us to remove chattering that arises because of the discretization of the underlying CT process. We show that the proposed algorithm converges to the SVM classifier over time-varying weight balanced directed graphs by using arguments from the matrix perturbation theory.
Classical distributed estimation scenarios typically assume timely and reliable exchanges of information over the sensor network. This paper, in contrast, considers single time-scale distributed estimation via a sensor network subject to transmission
time-delays. The proposed discrete-time networked estimator consists of two steps: (i) consensus on (delayed) a-priori estimates, and (ii) measurement update. The sensors only share their a-priori estimates with their out-neighbors over (possibly) time-delayed transmission links. The delays are assumed to be fixed over time, heterogeneous, and known. We assume distributed observability instead of local observability, which significantly reduces the communication/sensing loads on sensors. Using the notions of augmented matrices and Kronecker product, the convergence of the proposed estimator over strongly-connected networks is proved for a specific upper-bound on the time-delay.
This paper proposes networked dynamics to solve resource allocation problems over time-varying multi-agent networks. The state of each agent represents the amount of used resources (or produced utilities) while the total amount of resources is fixed.
The idea is to optimally allocate the resources among the group of agents by minimizing the overall cost function subject to fixed sum of resources. Each agents information is restricted to its own state and cost function and those of its immediate in-neighbors. This is motivated by distributed applications such as mobile edge-computing, economic dispatch over smart grids, and multi-agent coverage control. This work provides a fast convergent solution (in comparison with linear dynamics) while considering relaxed network connectivity with quantized communication links. The proposed dynamics reaches optimal solution over switching (possibly disconnected) undirected networks as far as their union over some bounded non-overlapping time-intervals has a spanning-tree. We prove feasibility of the solution, uniqueness of the optimal state, and convergence to the optimal value under the proposed dynamics, where the analysis is applicable to similar 1st-order allocation dynamics with strongly sign-preserving nonlinearities, such as actuator saturation.
In this paper, we study large-scale networks in terms of observability and controllability. In particular, we compare the number of unmatched nodes in two main types of Scale-Free (SF) networks: the Barab{a}si-Albert (BA) model and the Holme-Kim (HK)
model. Comparing the two models based on theory and simulation, we discuss the possible relation between clustering coefficient and the number of unmatched nodes. In this direction, we propose a new algorithm to reduce the number of unmatched nodes via link addition. The results are significant as one can reduce the number of unmatched nodes and therefore number of embedded sensors/actuators in, for example, an IoT network. This may significantly reduce the cost of controlling devices or monitoring cost in large-scale systems.
This paper considers the problem of simultaneous sensor fault detection, isolation, and networked estimation of linear full-rank dynamical systems. The proposed networked estimation is a variant of single time-scale protocol and is based on (i) conse
nsus on textit{a-priori} estimates and (ii) measurement innovation. The necessary connectivity condition on the sensor network and stabilizing block-diagonal gain matrix is derived based on our previous works. Considering additive faults in the presence of system and measurement noise, the estimation error at sensors is derived and proper residuals are defined for fault detection. Unlike many works in the literature, no simplifying upper-bound condition on the noise is considered and we assume Gaussian system/measurement noise. A probabilistic threshold is then defined for fault detection based on the estimation error covariance norm. Finally, a graph-theoretic sensor replacement scenario is proposed to recover possible loss of networked observability due to removing the faulty sensor. We examine the proposed fault detection and isolation scheme on an illustrative academic example to verify the results and make a comparison study with related literature.
Finite-time stability of networked control systems under Denial of Service (DoS) attacks are investigated in this paper, where the communication between the plant and the controller is compromised at some time intervals. Toward this goal, first an ev
ent-triggered mechanism based on the variation rate of the Lyapunov function is proposed such that the closed-loop system remains finite-time stable (FTS) and at the same time, the amount data exchange in the network is reduced. Next, the vulnerability of the proposed event-triggered finite-time controller in the presence of DoS attacks are evaluated and sufficient conditions on the DoS duration and frequency are obtained to assure the finite-time stability of the closed-loop system in the presence of DoS attack where no assumption on the DoS attack in terms of following a certain probabilistic or a well-structured periodic model is considered. Finally, the efficiency of the proposed approach is demonstrated through a simulation study.
Observability is a fundamental concept in system inference and estimation. This paper is focused on structural observability analysis of Cartesian product networks. Cartesian product networks emerge in variety of applications including in parallel an
d distributed systems. We provide a structural approach to extend the structural observability of the constituent networks (referred as the factor networks) to that of the Cartesian product network. The structural approach is based on graph theory and is generic. We introduce certain structures which are tightly related to structural observability of networks, namely parent Strongly-Connected-Component (parent SCC), parent node, and contractions. The results show that for particular type of networks (e.g. the networks containing contractions) the structural observability of the factor network can be recovered via Cartesian product. In other words, if one of the factor networks is structurally rank-deficient, using the other factor network containing a spanning cycle family, then the Cartesian product of the two nwtworks is structurally full-rank. We define certain network structures for structural observability recovery. On the other hand, we derive the number of observer nodes--the node whose state is measured by an output-- in the Cartesian product network based on the number of observer nodes in the factor networks. An example illustrates the graph-theoretic analysis in the paper.
