Ramu Sharat Chandra

Distributed Utilization Control for Real-Time Clusters with Load Balancing

Previous years have seen rapid growth of online services that rely on large-scale server clusters to handle high volume of requests. Such clusters must adaptively control the CPU utilizations of many processors in order to maintain desired soft real-time performance and prevent system overload in face of unpredictable workloads. This paper presents DUC-LB, a novel distributed utilization control algorithm for cluster-based soft real-time applications. Compared to earlier works on utilization control, a distinguishing feature of DUC-LB is its capability to handle system dynamics caused by load balancing, which is a common and essential component of most clusters today. Simulation results and control-theoretic analysis demonstrate that DUC-LB can provide robust utilization control and effective load balancing in large-scale clusters

Distributed control design with robustness to small time delays

Recent results by the authors have shown how to construct a class of structured controllers for large scale spatially interconnected systems via linear matrix inequalities. These controllers guarantee that the closed loop interconnected system is well-posed, stable and has H/sub /spl infin// norm less than unity. Of paramount importance in the control of interconnected systems is the requirement that the stability and performance of the controlled system be robust to arbitrarily small communication delays between subsystems; this amounts to a continuity property. In this paper, it is shown how to realize the structured controllers obtained from the linear matrix inequalities in order to ensure this continuity property for the closed loop system.

Optimal control of microgrids - algorithms and field implementation

A microgrid is a collection of distributed generation assets, storage devices and electrical and/or thermal loads connected to each other. In this paper, a generic model-predictive control algorithm for microgrids is presented. The algorithm has been implemented at Bella Coola, a remote community in British Columbia, Canada. The approach comprises two parts: unit commitment to decide the optimal set of distributed generators that must be switched on to meet predicted load requirements, and convex optimal control to minimize operational costs once the commitment is known. The unit commitment problem is recast as a 0-1 Knapsack problem and is solved via dynamic programming, while the optimal dispatch problem is posed as a sparse linear programming problem and solved via off-the-shelf software. Worst-case complexity and scalability considerations, and not optimality, often drive algorithm choice in industrial control settings; therefore, the solution proposed in this paper is efficient and can be rigorously bounded in terms of memory and run-time. Simulation results using real field data, practical considerations, and details of the implementation at Bella Coola are provided.

Antenna array synthesis with clusters of unmanned aerial vehicles

In this paper, we investigate the use of formations of unmanned aerial vehicles (UAVs) as phased antenna arrays. This will help improve communications with clusters of small unmanned aerial vehicles which are currently constrained by on-board power limitations. The problem of maximizing the power output from the array in the direction of the receiver is posed as an optimization problem which happens to be non-convex; a relaxation of this problem is then solved as a computationally tractable (convex) second order conic program (SOCP). The performance obtained by the simplified approach is then tested against rigorous numerical bounds obtained using semidefinite programming (SDP) duality theory; these bounds are of independent interest in antenna theory. In order to maintain the objective value close to the optimal when the vehicles deviate from their positions (due to wind gusts, for example), a linear control law is proposed. Simulation results are given to show the effectiveness of the proposed approaches