In this paper, we are interested in systems with multiple agents that wish to collaborate in order to accomplish a common task while a) agents have different information (decentralized information) and b) agents do not know the model of the system completely i.e., they may know the model partially or may not know it at all. The agents must learn the optimal strategies by interacting with their environment i.e., by decentralized Reinforcement Learning (RL). The presence of multiple agents with different information makes decentralized reinforcement learning conceptually more difficult than centralized reinforcement learning. In this paper, we develop a decentralized reinforcement learning algorithm that learns $ε$-team-optimal solution for partial history sharing information structure, which encompasses a large class of decentralized control systems including delayed sharing, control sharing, mean field sharing, etc. Our approach consists of two main steps. In the first step, we convert the decentralized control system to an equivalent centralized POMDP (Partially Observable Markov Decision Process) using an existing approach called common information approach. However, the resultant POMDP requires the complete knowledge of system model. To circumvent this requirement, in the second step, we introduce a new concept called "Incrementally Expanding Representation" using which we construct a finite-state RL algorithm whose approximation error converges to zero exponentially fast. We illustrate the proposed approach and verify it numerically by obtaining a decentralized Q-learning algorithm for two-user Multi Access Broadcast Channel (MABC) which is a benchmark example for decentralized control systems.
We consider optimal control of decentralized LQG problems for plants controlled by two players having asymmetric information sharing patterns between them. In one scenario, players are assumed to have a bidirectional error-free, unlimited rate communication channel with no delay in one direction and a unit delay in the other. In another scenario, the communication channel is assumed to be unidirectional with a unit delay. Delayed information sharing patterns in general do not admit linear optimal control laws and are thus difficult to control optimally. However, in these scenarios, we show tha…
This paper considers the optimal decentralized control for networked control systems (NCSs) with asymmetric partial information sharing between two controllers. In this NCSs model, the controller 2 (C2) shares its observations and part of its historical control inputs with the controller 1 (C1), whereas C2 cannot obtain the information of C1 due to network constraints. We present the optimal estimators for C1 and C2 respectively based on asymmetric observations. Since the information for C1 and C2 are asymmetric, the estimation error covariance (EEC) is coupled with the controller which means …
Optimality of Symmetric Independent Policies under Decentralized Mean-Field Information Sharing for Stochastic Teams and Equivalence with McKean-Vlasov Control of a Representative Agent
Royal Military College of Canada（加拿大皇家军事学院）；Bilkent University（比尔肯特大学）；Queen’s University（女王大学）
University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）
In decentralized stochastic control, standard approaches for sequential decision-making, e.g. dynamic programming, quickly become intractable due to the need to maintain a complex information state. Computational challenges are further compounded if agents do not possess complete model knowledge. In this paper, we take advantage of the fact that in many problems agents share some common information, or history, termed partial history sharing. Under this information structure the policy search space is greatly reduced. We propose a provably convergent, online tree-search based algorithm that do…
This paper presents an analysis and experimental validation of dispatchable virtual oscillator control (dVOC) for inverter-dominated power systems. dVOC is a promising decentralized control strategy that requires only local measurements to induce grid-forming behavior with programmable droop characteristics. It is dispatchable i.e., the inverters can vary their power generation via user-defined power set-points and guarantees strong stability. To verify its feasibility, a testbed comprising multiple dVOC-programmed inverters with transmission line impedances is designed. With an embedded synch…
High-voltage direct current (HVDC) is a commonly used technology for long-distance electric power transmission, mainly due to its low resistive losses. When connecting multiple HVDC lines into a multi-terminal HVDC (MTDC) system, several challenges arise. To ensure safe and efficient operation of MTDC systems, the voltage of all terminals need to be steered to within an operational range. In this paper we study the commonly used decentralized voltage droop controller, and show that it in general does not steer the voltages to within the operational range. We propose a decentralized PI controll…
This paper considers the problem of decentralized analysis and control synthesis to verify and ensure properties like stability and dissipativity of a large-scale networked system comprised of linear subsystems interconnected in an arbitrary topology. In particular, we design systematic networked system analysis and control synthesis processes that can be executed in a decentralized manner at the subsystem level with minimal information sharing among the subsystems. Compared to our most recent work on the same topic, we consider a substantially more generalized problem setup in this paper and …
We study strongly convex distributed optimization problems where a set of agents are interested in solving a separable optimization problem collaboratively. In this paper, we propose and study a two time-scale decentralized gradient descent algorithm for a broad class of lossy sharing of information over time-varying graphs. One time-scale fades out the (lossy) incoming information from neighboring agents, and one time-scale regulates the local loss functions' gradients. For strongly convex loss functions, with a proper choice of step-sizes, we show that the agents' estimates converge to the g…
The rapidly growing penetration of renewable energy resources brings unprecedented challenges to power distribution networks - management of a large population of grid-tied controllable devices encounters control scalability crises and potential end-user privacy breaches. Despite the importance, research on privacy preservation of distributed energy resource (DER) control in a fully scalable manner is lacked. To fill the gap, this paper designs a novel decentralized privacy-preserving DER control framework that 1) achieves control scalability over DER population and heterogeneity; 2) eliminate…
We consider cooperative multi-agent resource sharing problems over time-varying communication networks, where only local communications are allowed. The objective is to minimize the sum of agent-specific composite convex functions subject to a conic constraint that couples agents' decisions. We propose a distributed primal-dual algorithm DPDA-D to solve the saddle point formulation of the sharing problem on time-varying (un)directed communication networks; and we show that primal-dual iterate sequence converges to a point defined by a primal optimal solution and a consensual dual price for the…
We present and analyze a computational hybrid architecture for performing multi-agent optimization. The optimization problems under consideration have convex objective and constraint functions with mild smoothness conditions imposed on them. For such problems, we provide a primal-dual algorithm implemented in the hybrid architecture, which consists of a decentralized network of agents into which centralized information is occasionally injected, and we establish its convergence properties. To accomplish this, a central cloud computer aggregates global information, carries out computations of th…
Decentralized solutions to finite-sum minimization are of significant importance in many signal processing, control, and machine learning applications. In such settings, the data is distributed over a network of arbitrarily-connected nodes and raw data sharing is prohibitive often due to communication or privacy constraints. In this article, we review decentralized stochastic first-order optimization methods and illustrate some recent improvements based on gradient tracking and variance reduction, focusing particularly on smooth and strongly-convex objective functions. We provide intuitive ill…
Kuwait University（科威特大学）；Michigan State University（密歇根州立大学）；Ecole Polytechnique Federale de Lausanne（洛桑联邦理工学院）
This work studies multi-agent sharing optimization problems with the objective function being the sum of smooth local functions plus a convex (possibly non-smooth) function coupling all agents. This scenario arises in many machine learning and engineering applications, such as regression over distributed features and resource allocation. We reformulate this problem into an equivalent saddle-point problem, which is amenable to decentralized solutions. We then propose a proximal primal-dual algorithm and establish its linear convergence to the optimal solution when the local functions are strong…
A challenge in reinforcement learning (RL) is minimizing the cost of sampling associated with exploration. Distributed exploration reduces sampling complexity in multi-agent RL (MARL). We investigate the benefits to performance in MARL when exploration is fully decentralized. Specifically, we consider a class of online, episodic, tabular $Q$-learning problems under time-varying reward and transition dynamics, in which agents can communicate in a decentralized manner.We show that group performance, as measured by the bound on regret, can be significantly improved through communication when each…
Rutgers Business School（罗格斯大学商学院）；The Chinese University of Hong Kong（香港中文大学）；Florida State University（佛罗里达州立大学）
Stochastic gradient Langevin dynamics (SGLD) and stochastic gradient Hamiltonian Monte Carlo (SGHMC) are two popular Markov Chain Monte Carlo (MCMC) algorithms for Bayesian inference that can scale to large datasets, allowing to sample from the posterior distribution of the parameters of a statistical model given the input data and the prior distribution over the model parameters. However, these algorithms do not apply to the decentralized learning setting, when a network of agents are working collaboratively to learn the parameters of a statistical model without sharing their individual data …
