Bartłomiej Płaczek

A self-organizing system for urban traffic control based on predictive interval microscopic model

This paper introduces a self-organizing traffic signal system for an urban road network. The key elements of this system are agents that control traffic signals at intersections. Each agent uses an interval microscopic traffic model to predict effects of its possible control actions in a short time horizon. The executed control action is selected on the basis of predicted delay intervals. Since the prediction results are represented by intervals, the agents can recognize and suspend those control actions, whose positive effect on the performance of traffic control is uncertain. Evaluation of the proposed traffic control system was performed in a simulation environment. The simulation experiments have shown that the proposed approach results in an improved performance, particularly for non-uniform traffic streams.

Uncertainty-based information extraction in wireless sensor networks for control applications

Design of control applications over wireless sensor networks (WSNs) is a challenging issue due to the bandwidth-limited communication medium, energy constraints and real-time data delivery requirements. This paper introduces a new information extraction method for WSN-based control applications, which reduces the number of required data transmissions to save energy and avoid data congestion. According to the proposed approach, control applications recognize when new data readings have to be collected and determine sensor nodes that have to be activated on the basis of uncertainty analysis. Processing of the selectively collected input data is based on definition of information granules that describe state of the controlled system as well as performance of particular control decisions. This method was implemented for object tracking in WSNs. The task is to control movement of a mobile sink, which has to reach a target in the shortest possible time. Extensive simulation experiments were performed to compare performance of the proposed approach against state-of-the-art methods. Results of the experiments show that the presented information extraction method allows for substantial reduction in the amount of transmitted data with no significant negative effect on tracking performance.

Energy Aware Object Localization in Wireless Sensor Network Based on Wi-Fi Fingerprinting

The usage of GPS systems for indoor localization is limited, therefore multiple indirect localization techniques were proposed over the years. One of them is a localization method based on Wi-Fi (802.11) access point (AP) signal strength (RSSI) measurement. In this method, a RSSI map is constructed via Localization Fingerprinting (LF), which allows localizing object on the basis of a pattern similarity. The drawback of LF method is the need to create the RSSI map that is used as a training dataset. Therefore, in this study a Wireless Sensor Network (WSN) is used for this task. The introduced in this paper energy aware localization method allows to acquire the actual RSSI map or broadcast a localization signal, if there is not sufficient information to perform the localization by using nearby APs. To localize objects in a given cell, various classifiers were used and their localization accuracy was analyzed. Simulations were performed to compare the introduced solution with a state-of-the-art approach. The experimental results show that the proposed energy aware method extends the lifetime of WSN and improves the localization accuracy.

Period-aware local modelling and data selection for time series prediction

The introduced algorithm selects useful data for improved training of local models.A hybrid usefulness-related distance is proposed for training data selection.Data usefulness is evaluated by taking into account periodicity of time series.Autocorrelation function and Renyi entropy is used to reduce number of parameters.The proposed method offers lower prediction error than the state-of-the-art local and global models. The paper tackles with local models (LM) for periodical time series (TS) prediction. A novel prediction method is introduced, which achieves high prediction accuracy by extracting relevant data from historical TS for LMs training. According to the proposed method, the period of TS is determined by using autocorrelation function and moving average filter. A segment of relevant historical data is determined for each time step of the TS period. The data for LMs training are selected on the basis of the k-nearest neighbours approach with a new hybrid usefulness-related distance. The proposed definition of hybrid distance takes into account usefulness of data for making predictions at a given time step. During the training procedure, only the most informative lags are taken into account. The number of most informative lags is determined in accordance with the Kraskovs mutual information criteria. The proposed approach enables effective applications of various machine learning (ML) techniques for prediction making in expert and intelligent systems. Effectiveness of this approach was experimentally verified for three popular ML methods: neural network, support vector machine, and adaptive neuro-fuzzy inference system. The complexity of LMs was reduced by TS preprocessing and informative lags selection. Experiments on synthetic and real-world datasets, covering various application areas, confirm that the proposed period aware method can give better prediction accuracy than state-of-the-art global models and LMs. Moreover, the data selection reduces the size of training dataset. Hence, the LMs can be trained in a shorter time.

Fully connected neural networks ensemble with signal strength clustering for indoor localization in wireless sensor networks

The paper introduces a method which improves localization accuracy of the signal strength fingerprinting approach. According to the proposed method, entire localization area is divided into regions by clustering the fingerprint database. For each region a prototype of the received signal strength is determined and a dedicated artificial neural network (ANN) is trained by using only those fingerprints that belong to this region (cluster). Final estimation of the location is obtained by fusion of the coordinates delivered by selected ANNs. Sensor nodes have to store only the signal strength prototypes and synaptic weights of the ANNs in order to estimate their locations. This approach significantly reduces the amount of memory required to store a received signal strength map. Various ANN topologies were considered in this study. Improvement of the localization accuracy as well as speedup of learning process was achieved by employing fully connected neural networks. The proposed method was verified and compared against state-of-the-art localization approaches in real world indoor environment by using both stationary and mobile sensor nodes.

Data Suppression Algorithms for Surveillance Applications of Wireless Sensor and Actor Networks

This paper introduces algorithms for surveillance applications of wireless sensor and actor networks (WSANs) that reduce communication cost by suppressing unnecessary data transfers. The objective of the considered WSAN system is to capture and eliminate distributed targets in the shortest possible time. Computational experiments were performed to evaluate effectiveness of the proposed algorithms. The experimental results show that a considerable reduction of the communication costs together with a performance improvement of the WSAN system can be obtained by using the communication algorithms that are based on spatiotemporal and decision aware suppression methods.

Communication-Aware Algorithms for Target Tracking in Wireless Sensor Networks

This paper introduces algorithms for target tracking in wireless sensor networks (WSNs) that enable reduction of data communication cost. The objective of the considered problem is to control movement of a mobile sink which has to reach a moving target in the shortest possible time. Consumption of the WSN energy resources is reduced by transferring only necessary data readings (target positions) to the mobile sink. Simulations were performed to evaluate the proposed algorithms against existing methods. The experimental results confirm that the introduced tracking algorithms allow the data communication cost to be considerably reduced without significant increase in the amount of time that the sink needs to catch the target.

Road Traffic Monitoring System Based on Mobile Devices and Bluetooth Low Energy Beacons

The paper proposes a method, which utilizes mobile devices (smartphones) and Bluetooth beacons, to detect passing vehicles and recognize their classes. The traffic monitoring tasks are performed by analyzing strength of radio signal received by mobile devices from beacons that are placed on opposite sides of a road. This approach is suitable for crowd sourcing applications aimed at reducing travel time, congestion, and emissions. Advantages of the introduced method were demonstrated during experimental evaluation in real-traffic conditions. Results of the experimental evaluation confirm that the proposed solution is effective in detecting three classes of vehicles (personal cars, semitrucks, and trucks). Extensive experiments were conducted to test different classification approaches and data aggregation methods. In comparison with state-of-the-art RSSI-based vehicle detection methods, higher accuracy was achieved by introducing a dedicated ensemble of random forest classifiers with majority voting.

Wireless Network with Bluetooth Low Energy Beacons for Vehicle Detection and Classification

The paper proposes a hybrid wireless network, which can be installed on a roadside to detect passing vehicles and recognize their classes. The vehicle detection and classification tasks are performed by analyzing strength of a radio signal received from Bluetooth Low Energy beacons with the use of machine learning algorithms. The introduced system is cost-efficient, easy to install, and can be used for a long time without an external power source. An energy-aware algorithm is proposed, which uses a scheduling mechanism to manage wireless nodes that can act as BLE beacons (in low energy mode) or receivers. Results of experimental evaluation confirm that the proposed solution enables collection of accurate traffic data in real time and prolongs lifetime of battery-powered wireless nodes in the traffic monitoring system. The paper also discusses the applicability of various wireless communication technologies and the influence of wireless node location on vehicle detection accuracy.

Edge Real-Time Medical Data Segmentation for IoT Devices with Computational and Memory Constrains.

The Internet of Things (IoT) becomes very important tool for data gathering and management in many environments. The majority of dedicated solutions register data only at time of events, while in case of medical data full records for long time periods are usually needed. The precision of acquired data and the amount of data sent by sensor-equipped IoT devices has vital impact on lifetime of these devices. In case of solutions, where multiple sensors are available for single device with limited computation power and memory, the complex compression or transformation methods cannot be applied - especially in case of nano device injected to a body. Thus this paper is focused on linear complexity segmentation algorithms that can be used by the resource-limited devices. The state-of-art data segmentation methods are analysed and adapted for simple IoT devices. Two segmentation algorithms are proposed and tested on a real-world dataset collected from a prototype of the IoT device.