Now showing 1 - 10 of 57
  • Placeholder Image
    Publication
    Automated control of webserver performance in a cloud environment
    (01-12-2013)
    SaiKrishna, P. S.
    ;
    Cloud computing has been emerging as a new technology. In a distributed computing perspective cloud is similar to client-server services like web-based services and uses virtualized resources for execution. The widespread use of internet technology has focused attention on quality of service, especially the response time experienced by the end user. We demonstrate the performance degradation of traditional web hosting with time varying user requests directly affecting the response time. We also show how this issue could be addressed by a web-server hosted on a cloud using control algorithms for Load balancing and Elasticity control developed to maintain the desired response time within acceptable limit. Our experimental setup hosts a web server on an open source Eucalyptus cloud platform. To evaluate the control system performance we use the web server benchmarking tool called httperf and autobench for automating the process of benchmarking. © 2013 IEEE.
  • Placeholder Image
    Publication
    Welcome by the General Co-Chairs
    (01-12-2019)
    Vidyasagar, M.
    ;
  • Placeholder Image
    Publication
    Optimizing Average Controllability of Networked Systems
    (01-12-2019)
    Srighakollapu, Manikya Valli
    ;
    ;
    In this paper, we consider the controllability of a networked system where each node in the network has higher order linear time-invariant (LTI) dynamics. We employ a quantitative measure for controllability based on average controllability. We relate this metric to the network topology and the dynamics of individual subsystems that constitute each node of the networked system. Using this, we show that, under certain assumptions, the average controllability increases with increased interactions across subsystems in the network. Next, we consider the problem of identifying an appropriate network topology when there are constraints on the number of links that exist in the network. This problem is formulated as a set function optimization problem. We show that for our problem, this set function is a monotone increasing supermodular function. Since maximization of such a function with cardinality constraints is a hard problem, we implement a greedy heuristic to obtain a sub-optimal solution.
  • Placeholder Image
    Publication
    Stability analysis of constrained optimization dynamics via passivity techniques
    (01-01-2018)
    Kosaraju, K. C.
    ;
    Chinde, V.
    ;
    ;
    Kelkar, A.
    ;
    Singh, N. M.
    In this letter, we present passivity-based convergence analysis of continuous time primal-dual gradient method for convex optimization problems. We first show that a convex optimization problem with only affine equality constraints admits a Brayton Moser formulation. This observation leads to a new passivity property derived from a Krasovskii-type storage function. Second, the inequality constraints are modeled as a state dependent switching system. Using tools from hybrid systems theory, it is shown that each switching mode is passive and the passivity of the system is preserved under arbitrary switching. Finally, the two systems: 1) one derived from the Brayton Moser formulation and 2) the state dependent switching system, are interconnected in a power conserving way. The resulting trajectories of the overall system are shown to converge asymptotically, to the optimal solution of the convex optimization problem. The proposed methodology is applied to an energy management problem in buildings and simulations are provided for corroboration.
  • Placeholder Image
    Publication
    Tracking and stabilization of mechanical systems using reinforcement learning
    (07-03-2018)
    Bhuvaneswari, S.
    ;
    ;
    Ravindran, Balaraman
    ;
    The Interconnection and Damping Assignment Passivity Based Control (IDA-PBC) is a well-known method for control of complex physical systems in the port-Hamiltonian framework. Improvising on top of IDA-PBC which just focuses on stability, the memristive port-Hamiltonian control addresses performance concerns in the control task by providing a state-modulated damping term to IDA-PBC via a memristor element. The control way of implementing the memristive IDA-PBC first requires solving a set of Partial Differential Equations (PDEs) and then choosing a suitable memristance function for the system, out of which the former is a challenging math problem and the latter is a design problem. This paper employs reinforcement learning to learn the memristive IDA-PBC law and in the process, avoids the challenging task of solving PDEs, automates the design of the memristance function and also respects some physical system-level constraints which are not accounted for by the control way of solving IDA-PBC.
  • Placeholder Image
    Publication
    Time Series Analysis of Surface EMG Signal - Linear, Non Linear and Chaotic Approaches
    (14-05-2019)
    Singh, Moirangthem Sailash
    ;
    ;
    Talasila, Viswanath
    EMG data analysis forms a crucial part of clinical diagnosis in patients suffering from movement disorders. Any movement involves a synchronized use of agonist and antagonist muscles. In patients suffering from movement disorders, this coordination between different groups of muscles can be a problem. EMG data is used to study muscle activity; and the measurements are significantly noisy. To study a complex movement, one needs to derive useful models from the EMG data (for different muscle groups). In this paper we present a detailed analysis of surface EMG signals from the perspective of linear segmentation, non-linear and chaotic approaches. Comparisons of the linear regression models of surface EMG suggest the ARIMA model to be the best. A chaotic analysis of surface emg signal to determine the maximum lyaopunov exponent, the embedding dimension and time lag has been discussed in the later part of the paper. The EMG signal is found to follow high dimensional chaotic dynamics from the positive value of the Maximum Lyapunov Exponent.
  • Placeholder Image
    Publication
    Towards analog memristive controllers
    (01-01-2015)
    Saha, Gourav
    ;
    ;
    Khatavkar, Prathamesh
    Memristors, initially introduced in the 1970s, have received increased attention upon successful synthesis in 2008. Considerable work has been done on modeling and applications in specific areas, however, very little is known on the potential of memristors for control applications. Being nanoscopic variable resistors, it is intuitive to think of using them as a variable gain. The main contribution of this paper is the development of a memristive analog gain control framework and theoretic foundation of a control strategy which can be implemented using this framework. Analog memristive controllers may find applications in control of large array of miniaturized devices where robust and adaptive control is needed due to parameter uncertainty and ageing issues.
  • Placeholder Image
    Publication
    Design of Robust Gradient Method for Convex Optimization Problem using Biased Integral Quadratic Constraints
    (01-05-2020)
    Sawant, M.
    ;
    Moyalan, J.
    ;
    Sutavani, S.
    ;
    Sonam, K.
    ;
    ;
    Wagh, S.
    Gradient based methods are the most widely used algorithms for convex optimization problems due to their simplicity and reliability. Applications based on gradient methods works well in absence of any external factors (noise, precision error etc.) with appropriate step size. In presence of uncertainty though, the shortcomings of the gradient methods become obvious from increased convergence time or in severe cases even the failure of convergence depending upon the noise level and the step size. This paper tries to address the issue of achieving the optimal performance from the gradient methods in presence of uncertainty. A case study of gradient descent algorithm has been conducted in this paper. The robust convergence of gradient based convex optimization methods is transformed into an equivalent stabilization problem in a dynamical system framework. The paper proposes a novel characterization of optimization algorithms by adopting the biased integral quadratic constraint (BIQC) framework which is generally used for robust analysis of dynamical systems. Using this approach an algorithmic procedure is devised to perform dynamics computation of upper bound of the step size which guarantees robust convergence towards an equilibrium point even in presence of uncertainty.
  • Placeholder Image
    Publication
    End-to-end Autonomous Driving in Heterogeneous Traffic Scenario Using Deep Reinforcement Learning
    (01-01-2023)
    Chakraborty, Soumyajit
    ;
    Kumar, Subhadeep
    ;
    ;
    In this paper, we propose an end-to-end autonomous driving architecture for safe maneuvering in heterogeneous traffic using a reinforcement learning (RL) algorithm. Using the proposed architecture we develop an RL agent that can make driving decisions directly from the sensor data. We formulate the autonomous driving problem as a Markov Decision Process and propose different architectures using Deep Q -Networks for two types of sensor data - top view images of the autonomous vehicle (AV) and its surrounding vehicles and information on relative position and velocities of the surrounding vehicles w.r.t the AV. We consider a highway scenario and analyze the performance of the RL agent using the proposed architectures using the highway-env simulator. We compare the driving performance of the AV for both sensor types and discuss their efficacy under varying traffic densities.
  • Placeholder Image
    Publication
    Developmental maturation of causal signaling hubs in voluntary control of saccades and their functional controllability
    (20-10-2022)
    Zhang, Yuan
    ;
    Ryali, Srikanth
    ;
    Cai, Weidong
    ;
    Supekar, Kaustubh
    ;
    ;
    Padmanabhan, Aarthi
    ;
    Luna, Bea
    ;
    Menon, Vinod
    The ability to adaptively respond to behaviorally relevant cues in the environment, including voluntary control of automatic but inappropriate responses and deployment of a goal-relevant alternative response, undergoes significant maturation from childhood to adulthood. Importantly, the maturation of voluntary control processes influences the developmental trajectories of several key cognitive domains, including executive function and emotion regulation. Understanding the maturation of voluntary control is therefore of fundamental importance, but little is known about the underlying causal functional circuit mechanisms. Here, we use state-space and control-theoretic modeling to investigate the maturation of causal signaling mechanisms underlying voluntary control over saccades. We demonstrate that directed causal interactions in a canonical saccade network undergo significant maturation between childhood and adulthood. Crucially, we show that the frontal eye field (FEF) is an immature causal signaling hub in children during control over saccades. Using control-theoretic analysis, we then demonstrate that the saccade network is less controllable in children and that greater energy is required to drive FEF dynamics in children compared to adults. Our findings provide novel evidence that strengthening of causal signaling hubs and controllability of FEF are key mechanisms underlying age-related improvements in the ability to plan and execute voluntary control over saccades.