Distributed Adaptive Time-Varying Convex Optimization for Multi-agent Systems
This paper focus on the time-varying convex optimization problems with uncertain parameters. A new class of adaptive algorithms are proposed to solve time-varying convex optimization problems. Under the mild assumption of Hessian and partial derivative of the gradient with respect to time, the dependence on them is reduced through appropriate adaptive law design. By integrating the new adaptive optimization algorithm and a modified consensus algorithms, the time-varying optimization problems can be solved in a distributed manner. Then, they are extended from the agents with single-integrator dynamics to double-integrator dynamics, which describes more practical systems. As an example, the source seeking problem is used to verify the proposed design.
PDF Abstract