-
[show abstract]
[hide abstract]
ABSTRACT: Wireless sensor networks are deployed for the purpose of monitoring an area of interest. Even when the sensors are properly calibrated at the time of deployment, they develop drift in their readings leading to erroneous network inferences. Based on the assumption that neighbouring sensors have correlated measurements and that the instantiations of drifts in sensors are uncorrelated, the authors present a novel algorithm for detecting and correcting sensor measurement errors. The authors use statistical modelling rather than physical relations to model the spatio-temporal cross-correlations among sensors. This in principle makes the framework presented applicable to most sensing problems. Each sensor in the network trains a support vector regression algorithm on its neighbours' corrected readings to obtain a predicted value for its future measurements. This phase is referred to here as the training phase. In the running phase, the predicted measurements are used by each node, in a recursive decentralised fashion, to self-assess its measurement and to detect and correct its drift and random error using an unscented Kalman filter. No assumptions regarding the linearity of drift or the density (closeness) of sensor deployment are made. The authors also demonstrate using real data obtained from the Intel Berkeley Research Laboratory that the proposed algorithm successfully suppresses drifts developed in sensors and thereby prolongs the effective lifetime of the network.
Wireless Sensor Systems, IET. 07/2011;
-
[show abstract]
[hide abstract]
ABSTRACT: Automatic cell segmentation in phase contrast microscopy images play a very important role in the study the behavior of lymphocytes, such as cell motility, cell deformation, and cell population dynamics etc. In this paper, we have developed a set of algorithms for the microscopy image cell segmentation, in which three pairs of edge detection (Sobel, Prewitt and Laplace) based cell segmentation algorithms are developed in parallel to increase the probability of cell detection. Then, an hierarchical model is proposed and used in decision fusion that combine the three pair of detection results to increase the probability of final cell detection. After that, a false removal algorithm is proposed to remove false detections that may occur in the fusion process. The distance and watershed transforms have also been used to separate the connected cells. Experimental results have proved that these algorithms are pretty robust to variable microscopy image data, and variable cell densities, and with the proposed fusion and false removal algorithms, the cell detection rate has increased significantly to above 97% with the false detection rate about 7%.
Biomedical Engineering and Informatics, 2009. BMEI '09. 2nd International Conference on; 11/2009
-
[show abstract]
[hide abstract]
ABSTRACT: Random finite set approach is a mathematically rigorous framework for multi-target tracking. It provides a Bayesian recursion of multi-target distribution through finite set calculus. But practical implementation of multi-target posterior recursion is difficult because of its combinatorial nature. Probability hypothesis density (PHD) filter is an alternative to this problem where only the first order moment of the complete multi-target posterior is propagated in time. One of the suitable implementations of probability density filter is Gaussian mixture PHD filter. Parallel to this approach, several multi-target tracking algorithms are devised based on corresponding single target tracking algorithms. Joint integrated probabilistic data association is one of the most successful of such algorithms. This article shows that PHD filter recursion reduces to joint IPDA formalism under linear Gaussian assumptions.
Information Fusion, 2009. FUSION '09. 12th International Conference on; 08/2009
-
[show abstract]
[hide abstract]
ABSTRACT: Wireless sensor networks are deployed for the purpose of sensing and monitoring an area of interest. Sensor measurements in sensor networks usually suffer from both random errors (noise) and systematic errors (drift and bias). Even when the sensors are properly calibrated at the time of deployment, they develop errors in their readings leading to erroneous inferences to be made by the network. In this paper we present a novel algorithm for detecting and correcting sensor measurement errors by utilising the spatio-temporal correlation among the neighbouring sensors. The algorithm is designed for sparsely deployed wireless sensor networks. It can follow and correct both slowly and suddenly changing sensor measurements. As a result, the algorithm can adapt for under sampling the sensor measurements. Therefore, it allows for reducing the communication between sensors to maintain the calibration which leads to reducing the energy consumed from the batteries. The algorithm runs recursively and is totally decentralized. We demonstrate using real data obtained from the Intel Berkeley Laboratory that our algorithm successfully suppresses errors developed in sensors and thereby prolongs the effective lifetime of the network.
Information Fusion, 2009. FUSION '09. 12th International Conference on; 08/2009
-
[show abstract]
[hide abstract]
ABSTRACT: Automatic cell segmentation and tracking in optical microscope images plays a very important role in the study the behaviour of lymphocytes. The variable image contrasts, and especially variable cell densities are major factors to affect the successful cell detection rates. In this paper, two inner and outer cell contours edge detection based cell segmentation algorithms are proposed and used in parallel. Then a detection fusion algorithm is proposed to combine the two detection results and increase the probability of cell detection. Experimental results are used to demonstrate that these algorithms are robust to variations in both image contrast and cell densities. We show that the proposed fusion algorithm can increase cell detection rate significantly to above 90% with the false detection rate about 5%.
Multimedia Signal Processing, 2008 IEEE 10th Workshop on; 11/2008
-
[show abstract]
[hide abstract]
ABSTRACT: The purpose for wireless sensor networks is to deploy low cost sensors with sufficient computing and communication capabilities to support networked sensing applications. The emphasis on lower cost led to sensors that are less accurate and less reliable than their wired sensor counterparts. Sensors usually suffer from both random and systematic bias problems. Even when the sensors are properly calibrated at the time of their deployment, they develop drift in their readings leading to biased sensor measurements. The drift in this context is defined as a unidirectional long-term change in the sensor measurement. Assuming that neighboring sensors have correlated measurements and noting that the instantiation of drift in a sensor is uncorrelated with other sensors, we present the methodology for detecting and correcting sensors smooth and steep drifts. The methodology improves the reliability and the effective life of the network.
Communications, Computers and Applications, 2008. MIC-CCA 2008. Mosharaka International Conference on; 09/2008
-
[show abstract]
[hide abstract]
ABSTRACT: In this paper we extend our previously designed trust model in wireless sensor networks to include both; communication trust and data trust. It has been argued that if the overall trust is based on just one trust component, then it might mislead the network, and result in total breakdown of the network. Hence we introduce a new Bayesian network trust model to combine more than one trust component; communication trust and data trust in our case; to produce the overall trust. The model is robust and generic, which allows more trust components to be added to and/or removed from the model very easily.
Communications, Computers and Applications, 2008. MIC-CCA 2008. Mosharaka International Conference on; 09/2008
-
[show abstract]
[hide abstract]
ABSTRACT: In this paper we extend our previously designed trust model in wireless sensor networks to include both; communication trust and data trust. Trust management in wireless sensor networks is predominantly based on routing messages; whether the communication has happened or not (successful and unsuccessful transactions). The uniqueness of sensing data in wireless sensor networks introduces new challenges in calculating trust between nodes (data trust). If the overall trust is based on just the communication trust, it might mislead the network, that is; untrustworthy nodes in terms of sensed data can be classified as trusted nodes due to their communication capabilities. Hence we need to develop new trust models to address the issue of the actual sensed data. Here we are comparing the two trust models and proving that one model by itself is not enough to decide on the trustworthiness of a node, so new techniques are required to combine both data trust and communication trust.
Computer and Communication Engineering, 2008. ICCCE 2008. International Conference on; 06/2008
-
[show abstract]
[hide abstract]
ABSTRACT: This paper introduces a new trust model and a reputation system for wireless sensor networks based on a sensed continuous data. It establishes the continuous version of the beta reputation system and applied to binary events and presents a new Gaussian reputation system for sensor networks (GRSSN) . We introduce a theoretically sound Bayesian probabilistic approach for mixing second-hand information from neighbouring nodes with directly observed information.
Intelligent Sensors, Sensor Networks and Information, 2007. ISSNIP 2007. 3rd International Conference on; 01/2008
-
[show abstract]
[hide abstract]
ABSTRACT: The focus of wireless sensor networks is to develop low cost sensors with sufficient computing and communication capabilities to support networked sensing applications. The emphasis on lower cost led to sensors that are less accurate and less reliable than their wired sensor counterparts. Sensors usually suffer from both random and systematic (bias) problems. Even when the sensors are properly calibrated at the time of their deployment, they develop drift in their readings leading to biased sensor measurements. The drift in this context is defined as a unidirectional long-term change in the sensor measurement. Assuming that neighboring sensors have correlated measurements and noting that the instantiation of drift in a sensor is uncorrelated with other sensors and inspired by the resemblance of registration problem in radar target tracking with the bias error problem in sensor networks we devise a novel algorithm for detecting and correcting sensors drifts and show how it improves the reliability and the effective life of the network.
Information Fusion, 2007 10th International Conference on; 08/2007
-
[show abstract]
[hide abstract]
ABSTRACT: Intelligent video surveillance applications are frequently used nowadays. For such applications, background estimation becomes an important building block. In indoor environments, the switching of light takes place often. The previous approaches do not adapt quite well to the light switch. They take a long time to react to this problem. We attempt to tackle this problem by modeling it as a multiple model estimation problem. The well known IMM (interacting multiple model) estimator which has been successful in tracking maneuvering targets is used to tackle the light switch problem. The results show that it can easily detect the light switch with less error.
Information, Decision and Control, 2007. IDC '07; 03/2007
-
[show abstract]
[hide abstract]
ABSTRACT: Real time pedestrian flow information and the count of people in determined areas is essential for a multitude of management and monitoring functions. The range of applications is wide and the focus here is on the development of statistical methods for people volume estimation in buildings. To count people, visual information from surveillance cameras is used. This is the first of a series of reports, where the problem is divided in estimation classes of increasing difficulty. In the initial stage, probability models are derived for the basic counting scenario, using data gathered from a single camera. The methodology is extended to more complex problems, making use of several cameras covering the same field of view. The consideration of all possible classes of estimation problems leads to solutions that culminate in the assessment of the total number of people in a building.
Information, Decision and Control, 2007. IDC '07; 03/2007
-
[show abstract]
[hide abstract]
ABSTRACT: This paper focuses on the problem of color recognition from natural images under various illumination conditions. The most likely colors of the vehicle are estimated in different regions of the car image around the plate and the results from these regions are fused using the famous Bayes' rule. The application of the information fusion enhanced color recognition technology in stolen vehicle identification is also proposed in this paper. The basic approach is based on jointly estimating the license plate characters and other features of the car like its color and type and matching them with the registered car databases. If there is an inconsistency, the proposed method classifies the vehicle under observation as possibly stolen.
Information, Decision and Control, 2007. IDC '07; 03/2007
-
[show abstract]
[hide abstract]
ABSTRACT: We have defined a new paradigm of robot navigation, by using sensors in the environment instead of onboard robotic platform. In this scenario, determining the output of the most relevant and validated sensor is of crucial importance when heterogeneous sensors are available for measuring a given process. We are using an IEEE 1451 TEDS (transducer electronic data sheets) compliant sensor model proposed earlier for heterogeneous sensor networks. The presented method computes at each time step the usefulness of sensor information from discovered sensors, based upon Kullback-Leibler divergence between the sources of information. The robotic application consequently can obtain the relevance of the sensor output and the robot can update its estimate of the environment. We present results of a real time implementation of heterogeneous sensor networks using distributed multi-sensing 3D real-time robotics software player/Gazebo on a simulated pioneer robot. The results show that the proposed model can be utilized in the real-time scenario and can help reduce the computational cost of a system which occurs due to lack of discriminatory information
Robotics, Automation and Mechatronics, 2006 IEEE Conference on; 01/2007
-
[show abstract]
[hide abstract]
ABSTRACT: This paper presents a fixed lag smoothing algorithm for target tracking in clutter. The proposed algorithm is based on integrated probabilistic data association (IPDA) approach. The algorithm runs two filters for each track-one in forward direction (as standard IPDA) and the other in backward direction. A standard fusion of both the target existence probability and state estimates of both the filters at a fixed time lag yields the smoothed probability densities of both of them. The paper also presents the refined target dynamics and target existence transition model for backward running IPDA filter. Simulation results are also presented to compare the performance (in terms of true track detection, false track discrimination) of the proposed IPDA smoother with that of augmented state IPDA filter and standard IPDA filter
Information Fusion, 2006 9th International Conference on; 08/2006
-
[show abstract]
[hide abstract]
ABSTRACT: Electrical market often demands accurate forecasting of electrical load for planning and operation of the power infrastructure. Current models can forecast load from half hour up to 24 hours and are based on aggregate temperature for the entire day. Although these models work very well, they do not consider the intermediate real time information between time intervals to forecast load which introduce many uncertainties pertaining to factors such as climatic conditions, geographic locations etc. Furthermore, such intermediate real time information is costly and difficult to obtain. With the aid of distributed sensor networks, real time information can easily be obtained which can lead to precise planning and operation of power systems. Such information can easily improve electrical load forecasting and reduce uncertainty which can have a direct impact on the customer. We propose new and improved models for electricity load forecasting by incorporating real-time weather (temperature) information arising from the low-cost distributed sensor networks
Intelligent Sensing and Information Processing, 2005. ICISIP 2005. Third International Conference on; 01/2006
-
[show abstract]
[hide abstract]
ABSTRACT: Determining the output of the most relevant sensor is of crucial importance when heterogeneous sensors are available for measuring a given process in an environment. In this paper, we describe an IEEE 1451 TEDS (transducer electronic data sheets) compliant sensor model for heterogeneous sensor networks. The proposed model uses the relevance feedback method to understand the context of a sensor learning application. We present results of a real time implementation of heterogeneous sensor networks using distributed multi-sensing 3D real-time robotics software player/gazebo on an autonomous mobile robot's navigation problem. The results show that the proposed model can be utilised in the real-time scenario and can help reduce the computational cost of a system
Computational Advances in Multi-Sensor Adaptive Processing, 2005 1st IEEE International Workshop on; 01/2006
-
[show abstract]
[hide abstract]
ABSTRACT: Oppurtunistic Information Fusion (OIF) is introduced to enable the same sensors to provide data for multiple applications. Sensor location plays a crucial rule to get the maximum amount of useful information. This paper examines the optimal placement of cameras for a Networked Sensing Systems (NSS) that are designed to monitor a pre defined region to have as much coverage as possible with the purpose of serving multiple applications. This can be rephrased as a camera location optimization problem with multiple objective functions. Multi-Objective Genetic Algorithms (MOGA) is used with camera coverage as the two objective functions to be maximised
Intelligent Sensors, Sensor Networks and Information Processing Conference, 2005. Proceedings of the 2005 International Conference on; 01/2006
-
[show abstract]
[hide abstract]
ABSTRACT: Wireless sensors became smaller and cheaper in the recent years. Applications with thousands of nodes for tracking and monitoring are now feasible. Many of them require the knowledge about the locations of the sensors. The process of localization depends on estimating the position of the features within the environment. This paper proposes a novel algorithm to localize the sensors in any environment. Unlike any other technique like conventional SLAM, this new method does not require any architecture involving non-active beacons or mobile agents, such as robot. The new algorithm is presented that uses the sensors which act as active beacons themselves. The localization uses a least squares estimation (LSE), which processes the measured distances between the sensors. The distance measured was calculated based on received signal strength indicators (RSSI). Experiments were carried out in Mica2 Motes sensor networks. The estimated locations in the experiments were less then one meter away from the true locations which is an error of less than 10% of the transmission range from the sensors.
Intelligent Sensors, Sensor Networks and Information Processing Conference, 2005. Proceedings of the 2005 International Conference on; 01/2006
-
[show abstract]
[hide abstract]
ABSTRACT: Traditionally, information fusion systems assume that the information is gathered from known sensors over proprietary communication networks and fuse using fixed rules of information fusion and designated computing and communication resources. Emerging technologies like wireless sensor networks, TEDS enabled legacy sensors, ubiquitous computing devices and all IP next generation networks are challenging the rationale of conventional information fusion systems. The technology has matured to a point where it is reasonable to discover sensors based on the context, establish relevance, query for appropriate data, and fuse it using the most appropriate fusion rule, using ubiquitous computing and communication environment in an opportunistic manner. We define such fusion systems as opportunistic information fusion systems. In this paper we introduce this new paradigm for information fusion and identify plausible approaches and challenges to design, develop and deploy the proposed next generation opportunistic information fusion systems.
Information Fusion, 2005 8th International Conference on; 08/2005
