Phani Chavali

A Distributed Algorithm of Appliance Scheduling for Home Energy Management System

Demand side management encourages the users in a smart grid to shift their electricity consumption in response to varying electricity prices. In this paper, we propose a distributed framework for the demand response based on cost minimization. Each user in the system will find an optimal start time and operating mode for the appliances in response to the varying electricity prices. We model the cost function for each user and the constraints for the appliances. We then propose an approximate greedy iterative algorithm that can be employed by each user to schedule appliances. In the proposed algorithm, each user requires only the knowledge of the price of the electricity, which depends on the aggregated load of other users, instead of the load profiles of individual users. In order for the users to coordinate with each other, we introduce a penalty term in the cost function, which penalizes large changes in the scheduling between successive iterations. Numerical simulations show that our optimization method will result in lower cost for the consumers, lower generation costs for the utility companies, lower peak load, and lower load fluctuations.

Scheduling and Power Allocation in a Cognitive Radar Network for Multiple-Target Tracking

We propose a cognitive radar network (CRN) system for the joint estimation of the target state comprising the positions and velocities of multiple targets, and the channel state comprising the propagation conditions of an urban transmission channel. We develop a measurement model for the received signal by considering a finite-dimensional representation of the time-varying system function which characterizes the urban transmission channel. We employ sequential Bayesian filtering at the receiver to estimate the target and the channel state. We propose a hybrid Bayesian filter that operates by partitioning the state space into smaller subspaces and thereby reducing the complexity involved with high-dimensional state space. The feedback loop that embodies the radar environment and the receiver enables the transmitter to employ approximate greedy programming to find a suitable subset of antennas to be employed in each tracking interval, as well as the power transmitted by these antennas. We compute the posterior Cramer-Rao bound (PCRB) on the estimates of the target state and the channel state and use it as an optimization criterion for the antenna selection and power allocation algorithms. We use several numerical examples to demonstrate the performance of the proposed system.

Parallel Load Schedule Optimization With Renewable Distributed Generators in Smart Grids

We propose a framework for demand response in smart grids that integrates renewable distributed generators (DGs). In this model, some users have DGs and can generate part of their electricity. They can also sell extra generation to the utility company. The goal is to optimize the load schedule of users to minimize the utility companys cost and user payments, while considering user satisfaction. We employ a parallel autonomous optimization scheme, where each user requires only the knowledge of the aggregated load of other users, instead of the load profiles of individual users. All the users can execute distributed optimization simultaneously. The distributed optimization is coordinated through a soft constraint on changes of load schedules between iterations. Numerical examples show that our method can significantly reduce the peak-hour load and costs to the utility and users. Since the autonomous user optimization is executed in parallel, our method also significantly decreases the computation time and communication costs.

Distributed Power System State Estimation Using Factor Graphs

We propose a distributed and a dynamic algorithm for a power system state estimation. We model the dependencies among the state vectors of neighboring areas and among the state vectors at different times using a factor graph. We then derive message update rules and use these rules to implement a sum-product message passing algorithm on the graph. In message passing, neighboring areas exchange messages which represent their beliefs about the unknown state vectors based on all the related measurements. These beliefs are then used to compute the posterior distribution of the power system state. In our paper, we represent the messages using a particle based approximation. Such a particle-based representation provides a simple and a computationally feasible method to update the messages in each iteration. Further, it allows us to model the nonlinearities present in the power system, and hence leads to a better performance accuracy compared with the traditional methods that use linear models. We show the accuracy of the proposed method via numerical simulations using the IEEE 14 and 118 bus systems as examples.

Managing Multi-Modal Sensor Networks Using Price Theory

We propose a unified framework for sensor management in multi-modal sensor networks, which is inspired by the trading behavior of economic agents in commercial markets. Each sensor node (SN) acts as a seller who wants to sell the data it collects, to the sensor network manager (SM) who acts as a buyer. The resources and the data are priced by looking to balance global supply and demand, with the SN required to purchase resources for producing the data, and the SM required to purchase data to accomplish his tasks. We model this interaction as a double sided market, with both consumers and producers, and propose an iterative double auction mechanism for computing the equilibrium of such a market. We relate the equilibrium point to the solutions of sensor selection (SS), resource allocation (RA), and data fusion (DF) problems, which constitute the sensor management. The proposed framework will enable the system to determine the kind and the amount of data that should be produced, and to combine the data that is produced at each SN. To illustrate this framework, we consider the problem of multiple-target tracking as an example. Numerical examples demonstrate the effectiveness of the proposed method, and show that appropriate sensor management will result in an accurate estimate of the number of targets in the scene, higher correct identifications of the targets, and a lower mean-squared error in the estimates of their positions and velocities.

Cognitive radar for target tracking in multipath scenarios

In this paper, we propose a cognitive radar system for target tracking in the presence of multipath reflections. We exploit the inherent spatial diversity offered by the multipath environment by constructing a new measurement vector, which we refer to as a virtual measurement vector. We employ broadband Orthogonal Frequency Division Multiplexing (OFDM) signalling at the transmitter and implement adaptive waveform design by minimizing the posterior Cramér Rao bound (PCRB) on the target state estimates to find the optimal weights to be transmitted on each subcarrier bin of the OFDM signal. We demonstrate with numerical simulations that the mean square error in the case of a cognitive radar is significantly lower than the mean square error in the case of a standard radar.

Concurrent Particle Filtering and Data Association Using Game Theory for Tracking Multiple Maneuvering Targets

We propose a particle filtering technique to track multiple maneuvering targets in the presence of clutter. We treat data association and state estimation, which are the two important sub-problems in tracking, as separate problems. We develop a game-theoretic framework to solve the data association, in which we model each tracker as a player and the set of measurements as strategies. We develop utility functions for each player, and then use a regret-based learning algorithm to find the equilibrium of this game. The game-theoretic approach allows us to associate measurements to all the targets simultaneously. Further, in contrast to the traditional Monte-Carlo data association algorithms that use samples of the association vector obtained from a proposal distribution, our method finds the association in a deterministic fashion. We then use Monte-Carlo sampling on the reduced dimensional state of each target, independently, and thereby mitigate the curse-of-dimensionality problem that is known to occur in particle filtering. We provide a number of numerical results to demonstrate the performance of our proposed filtering algorithm.

Parallel autonomous optimization of demand response with renewable distributed generators

We propose a framework for demand response in smart grids that integrate renewable distributed generators (DGs). In this framework, some users have DGs and can generate part of their electricity. They can also sell extra generation to the utility company. The goal is to optimize the load schedule of users to minimize the utility companys cost and user payments. We employ parallel autonomous optimization, where each user requires only knowledge of the aggregated load of other users instead of the load profiles of individual users, and can execute distributed optimization simultaneously. We performed numerical examples to validate our algorithm. The results show that our method can significantly lower peak hour load and reduce the costs to users and the utility. Since the autonomous user optimizations are executed in parallel, our method also dramatically decreases the computation time, management complexity, and communication costs.

Hierarchical particle filtering for multi-modal data fusion with application to multiple-target tracking

We propose a sequential and hierarchical Monte Carlo Bayesian framework for state estimation using multi-modal data. The proposed hierarchical particle filter (HPF) estimates the global filtered posterior density of the unknown state in multiple stages, by partitioning the state space and the measurement space into lower dimensional subspaces. At each stage, we find an estimate of one partition using the measurements from the corresponding partition, and the information from the previous stages. We demonstrate the proposed framework for joint initiation, termination and tracking of multiple targets using multi-modal sensors. Here, the multi-modal data consists of the measurements collected from a radar, an infrared camera and a human scout. We compare the performance of the proposed HPF with the performance of a standard particle filter that uses linear opinion (SPF-LO), independent opinion (SPF-IO), and independent likelihood (SPF-IL) for data fusion. The results show that HPF improves the robustness of the tracking system in handling the initiation and termination of targets and provides a lower mean-squared error (RMSE) in the position estimates of the targets that maintain their tracks. The RMSE in the velocity estimates using the HPF was similar to the RMSE obtained using SPF based methods.

A low-complexity sparsity-based multi-target tracking algorithm for urban environments

In this paper, we propose a low-complexity sparsity based multi-target tracking algorithm. We develop a finite dimensional representation of the received signal when the radar is operating in an urban environment. The dimensionality of the representation denotes the extra degrees of freedom that an urban environment offers. We employ spread-spectrum signaling to exploit the full diversity offered by the environment. We then develop a block-sparse measurement model by discretizing the delay-Doppler plane and prove that the dictionary of the block-sparse model exhibits a special structure under spread-spectrum signaling. This structure enables an efficient support recovery of the sparse vector, by projecting the measurement vector on the row space of the dictionary. Numerical simulations show that our tracking procedure takes significantly less time, while giving good tracking performance.