Department Publications

In this paper, we consider a cellular network operating according to the classical opportunistic beamforming and scheduling scheme. In a multi-cell network, interference is the main bottleneck on the communication rate, compared to noise. In this paper, we focus on deriving the ergodic downlink sum-rate scaling law for the system of interest considering an interference-limited regime (i.e., the high SNR region). More specifically, through deriving the moment generating function of the sum-rate, we show that the ergodic downlink sum-rate scales logarithmically as the number of users grows large. This result is different from the well-known double logarithmic sum-rate growth with increasing number of users due to the interference-limited assumption. Finally, numerical results are also presented to validate our analysis and provide further insights.

Transferring computational tasks from user-worn devices to everyday objects allows users to freely focus on their regular non-computing tasks. Identifying micro-activities (short-repetitive-activities that compose the macro-level behavior) enables the understanding of subtle behavioral changes and providing just-in-time information without explicit user input. In this paper, we propose the concept of micro-activity recognition of augmented-everyday-objects and evaluate the applicability of machine learning algorithms that are previously used for macro-level activity recognition. We outline a few proof-of-concept application scenarios that provide micro-activity-aware just-in-time information.

Localization is a significant consideration in a Wireless Sensor Network (WSN), as it is essential to correlate sensor information with the geographic location it originates from. The DV-Hop algorithm, being a range-free, distributed algorithm, is an attractive option in this regard, due to its simplicity. In this paper, we present two algorithms to improve the localization accuracy of the DV-Hop algorithm. The Hop Size derived through a distributed fashion by the DV-Hop algorithm is optimized using the knowledge of the position of the anchor nodes. The locations of the unknown nodes are then estimated using the optimum hop size. We examine the performance of the proposed algorithms through simulations in environments with varying node numbers, anchor ratios and radio range. The results show that the proposed algorithms have improved performance in localization accuracy and stability compared to the original …

The downlink performance of a dual-cell relay enhanced wireless network is investigated. A single relay is shared by the two base stations in the system. The relay is assumed to be placed in a region covered by both base stations. The relay station is operating in the decode-and-forward mode. Closed-form approximations which follow the exact results extremely closely are derived for the outage probability and the ergodic rate of the system.

Complex Event Processing (CEP) is an emerging field in high performance computing paradigm where real time (low latency) computing capability is expected over big data processing (high throughput). Significant number of software architectures have been developed to improve the throughput while reduce the latency but maintaining of the both aspects reaches the limits of the software platforms. This paper proposes a novel custom hardware accelerator architecture framework for CEP in big data domain. The proposed design improves the throughput performance more than 10 times over the software counterpart while keeping the latency value at less than 100 nano seconds. Same Structured Query Language(SQL) type queries used in reference software architecture were used to improve the flexibility. A query compiler based on the same query language grammar was designed to convert the queries in to …

Action recognition in a video plays an important role in computer vision and finds many applications in areas such as surveillance, sports, and elderly monitoring. Existing methods mostly rely on stationary backgrounds. Action recognition in dynamic backgrounds typically requires standard preprocessing steps such as motion compensation, background modeling, moving object detection and object recognition. The errors of the motion compensation step and background modelling increase the mis-detections. Therefore action recognition in dynamic background is challenging. In this paper, we use a combination of pose characterized by a silhouette and optic flows synthesized into a histogram. This enables us to classify the movement of the actor versus movement of the background. We use four background models to extract the silhouette from the frame. We use SVM to recognize actions, according to several …

The biological world is an ideal place for seeking inspiration for developing mathematical optimization algorithms. In this paper we propose two hybrid stochastic optimization algorithms that bear resemblance to the sexual reproduction cycle of Jellyfish and asexual reproductive cycle of species of Hydra. The performance of these two algorithms are investigated against other common optimization algorithms on a set of benchmark optimization problems. The results show that the proposed algorithms perform well.

This paper studies a cellular network in which base station (BS) locations are modeled using a homogenous spatial Poisson point process, and user locations are modeled using an arbitrary stationary homogenous point process. The objective is to maximize the Coverage-Intensity Product (CIP), which represents the average number of users covered per unit area, under constraints on transmit power, BS intensity and transmit power density (per unit area). Expressions for the CIP are obtained considering two user coverage models that differ from each other depending on the BS-user association. Using these expressions, solutions for the optimization problem are provided. All these results are presented for general bounded path loss models that satisfy some mild conditions. The results are also illustrated using a specific bounded path loss model to provide further insights.

Image Deblurring algorithms have been evolving over many decades. Although there are many algorithms available today that produce reasonably good results, their speed of execution make them less appealing for many real time applications. Thus we address this issue by improving speed of execution of an existing algorithm using parallelization and other optimizations. The algorithm we selected was published by Fergus et al. in 2006, for which a 10x speed up improvement was achieved. Many other authors have since followed his approach and have come up with different variants of the solutions to which our implementation could be readily extended to improve the run-time.

In this paper a transformerless voltage source sine wave inverter is proposed. A previously developed tri-state boost converter is utilized for the stepping up operation. In this topology the dynamic response of the inverter is increased, by avoiding the right hand plane zero in the converter's small-signal control-to-output transfer function when operating in the continuous conduction mode. The reference voltage for the tri-state boost converter is determined from the measured output voltage of the sine wave inverter and the calculated instantaneous value of the sine wave. The inverter with the tri-state boost converter and the inverter with a classical boost converter with voltage mode control and current mode control are compared for the dynamic response and efficiency. The analytical work of the design has been verified using a simulation. Furthermore, the boost converter is implemented in hardware to verify the boost …