Samir Sahyoun

Dynamic plume tracking using mobile sensors

Plume localization and prediction using mobile sensors is the primary contribution of this paper. Plume concentration values, measured by chemical sensors at different locations, are used to estimate the source of the plume. This is achieved by employing a stochastic approximation technique to localize the source and compare its performance with the nonlinear least squares method. The source location is then used as the initial estimate for the boundary tracking problem. Sensor measurements are used to estimate the parameters and states of the state space model of the dynamics of the plume boundary. The predicted locations are the reference inputs for the LQR controller. Measurements at the new locations (after the correction of the prediction error) are added to the set of data to refine the next prediction step. Interpolation, using the sensors locations, is used to approximate the boundary shape. An illustrative two-dimensional example is provided.

Source Localization using Stochastic Approximation and Least Squares Methods

This paper presents two approaches to locate the source of a chemical plume; Nonlinear Least Squares and Stochastic Approximation (SA) algorithms. Concentration levels of the chemical measured by special sensors are used to locate this source. Non‐linear Least Squares technique is applied at different noise levels and compared with the localization using SA. For a noise corrupted data collected from a distributed set of chemical sensors, we show that SA methods are more efficient than Least Squares method. SA methods are often better at coping with noisy input information than other search methods.

Reduced order modeling for fluid flows subject to quadratic type nonlinearities

Explicit model reduction for nonlinear systems with no prior information about the type of nonlinearity involved is difficult and challenging. It is easier to reduce nonlinear systems which nonlinearity is known. In this paper we introduce two nonlinear model reduction techniques for quadratic nonlinear systems. The first technique is nonlinear balanced truncation. The Controlability and observability gramians are computed by solving the Hamilton Jacobi equations and then used to find the transformation function to get the nonlinear balanced truncated system. The second technique is using Arnoldi algorithm. We apply both techniques to a practical nonlinear quadratic system which is the two-dimensional Burgers equation problem of a fluid passing an obstacle.

Distributed Stochastic Power Control for Time-Varying Long-Term and Short-Term Fading Wireless Networks

In this paper, new time-varying wireless channel models that capture both the space and time variations of long- term and short-term fading wireless networks are developed. The proposed models are based on stochastic differential equations. These models are more realistic than the static ones usually encountered in the literature. Moreover, optimal power control algorithms based on the new models are proposed. A centralized power control algorithm is shown to reduce to a simple linear programming problem if predictable power control strategies are used. In addition, an iterative distributed stochastic power control algorithm is used to solve for the optimization problem using stochastic approximations. The latter solely requires each mobile to know its received signal to interference ratio unlike common stochastic algorithms found in the literature. Numerical results show that the proposed distributed stochastic power control algorithm under the new time-varying channels provides better power stability and consumption than the deterministic ones.

Reduced-order spectral data modeling based on local proper orthogonal decomposition

Spectral imaging typically generates a large amount of high-dimensional data that are acquired in different sub-bands for each spatial location of interest. The high dimensionality of spectral data imposes limitations on numerical analysis. As such, there is an emerging demand for robust data compression techniques with loss of less relevant information to manage real spectral data. In this paper, we describe a reduced-order data modeling technique based on local proper orthogonal decomposition (POD) in order to compute low-dimensional models by projecting high-dimensional clusters onto subspaces spanned by local reduced-order bases. We refer to the proposed method as the local-based approach because POD finds locally optimal solutions on each group split by k-means clustering. Experimental results are reported on three public domain databases and an in-house database. Comparisons with three leading spectral recovery techniques, three decomposition techniques used for hyperspectral imaging, and two baseline techniques show that the proposed method leads to promising improvement on spectral and colorimetric accuracy corresponding to the reconstructed spectral reflectance.

Control and room temperature optimization of energy efficient buildings

The building sector consumes a large part of the energy used in the United States and is responsible for nearly 40% of greenhouse gas emissions. It is therefore economically and environmentally important to reduce the building energy consumption to realize massive energy savings. In this paper, a method to control room temperature in buildings is proposed. The approach is based on a distributed parameter model represented by a three dimensional (3D) heat equation in a room with heater/cooler located at ceiling. The latter is resolved using finite element methods, and results in a model for room temperature with thousands of states. The latter is not amenable to control design. A reduced order model of only few states is then derived using Proper Orthogonal Decomposition (POD). A Linear Quadratic Regulator (LQR) is computed based on the reduced model, and applied to the full order model to control room temperature.

Improved localization in GPS-denied environments using an autoregressive model and a generalized linear model

The Theater Positioning System (TPS), which can perform in GPS-denied environments and can work with, or independently of, GPS systems, was presented in [1]. The principal difficulty in optimally combining this new system and GPS is introduced by the environment, which may impart somewhat unpredictable transmission delays to the signal and thus results in less accurate performance when TPS works unaided in the environment while GPS is unavailable. In this paper, we propose two methods-an autoregressive process and a generalized linear model to model the transmission delays generated in TPS signal propagation. Using those, the unknown signal delays can be predicted and thus can be employed in the subsequent localization process. Numerical examples are provided to illustrate the performance of both methods proposed in this paper.

Orthogonal Locality Preserving model reduction and flow separation control for incompressible Navier Stokes equations

Orthogonal Locality Preserving Projections (OLPP) are locally linear but globally nonlinear projection maps that arise by solving a variational problem, which optimally preserves the neighborhood structure of the state dynamics. OLPP projects the data in a way that preserves a certain affinity graph where graph nodes are the snapshots from numerical solutions. The edges can be defined by talking some number of nearest neighbor nodes to every state or alternatively by including all neighbors within a defined radius around the states. In this paper we design an OLPP POD model order reduction technique for an incompressible Navier-Stokes system that governs the velocity and pressure behavior of a fluid passing through the NACA 0015 airfoil with periodic boundary conditions. We first compute the optimal POD basis functions and show the reduced order system, then we compute the OLPP reduced system. Finally we present the flow separation problem.