Undetected leaks in water distribution networks are a significant problem both economically and environmentally. Across Australia 12% of water is estimated to be lost through leaks and the annual cost to water utilities worldwide is US$14 billion. A sensor network that measures water flow in the pipes can be used to predict the location and size of leaks. Recent advances in sensor technology and lower costs mean that large scale sensor networks may soon be an economic choice for solving the leak detection problem. This paper presents a sensor network design method that generates human-readable rules for leak detection. Additionally, for a given network and range of operating scenarios, it discovers the best locations for flow sensors. The method is demonstrated to make acceptably accurate predictions under real-world conditions of uncertain measurements. It also allows trade-offs to be made between minimising the costs of installing and maintaining sensors and maximising prediction accuracy. For example, in some cases sufficiently accurate predictions can be made using sensors on only half the pipes.
The Internet of Things (IoT) is opening the doors to many new devices and applications. Such an increase in the variety of applications requires reconfigurable, flexible and expandable hardware for fabrication and development cost reduction. This has been achieved for the digital part with devices like Arduino. However, the sensor readout Analog-Front-End (AFE) circuits are mainly designed for a specific sensor type or application. Such an approach would be feasible for the current small number of applications and sensors used. However, it will increase cost drastically as the variety and number of applications and sensors are increased. Moreover, flexibility and expandability of the system will be limited. Therefore, a universal sensor platform that can be reconfigured to adapt to various sensors and applications is needed. Moreover, an array of such circuit can be made with the same sensor to increase measurement accuracy and reliability. It can also be used to integrate heterogeneous sensors for increasing the flexibility of the system, which will make the system adaptable to many applications through only activating the desired sensors. In this paper, an 8-mode reconfigurable sensor readout AFE with offset-cancellation-resolution enhancing scheme is proposed to serve as a step towards a universal sensor interface. The proposed AFE can be reconfigured to interface resistive, capacitive, current producing, and voltage producing sensors through direct or capacitive connection to its terminals. The proposed system is fabricated in 180nm CMOS process and has successfully measured the four types of sensor outputs. It has also been interfaced to Arduino board to allow easy interfacing of various sensors. Therefore, the proposed work can be used as general purpose AFE resulting in manufacturing and development cost reduction and increased flexibility and expandability.
Acoustic classification of anurans (frogs) has received increasing attention for its promising application in biological and environment studies. In this study, a novel feature extraction method for frog call classification is presented based on the analysis of spectrograms. The frog calls are first automatically segmented into syllables. Then, spectral peak tracks are extracted to separate desired signal (frog calls) from background noise. The spectral peak tracks are used to extract various syllable features, including: syllable duration, dominant frequency, oscillation rate, frequency modulation, and energy modulation. Finally, a k-nearest neighbor classifier is used for classifying frog calls based on the results of principal component analysis. The experiment results show that syllable features can achieve an average classification accuracy of 90.5% which outperforms Mel-frequency cepstral coefficients features (79.0%).
Our proposed scour monitoring system utilized low-cost commercial sensors, hall-effect sensors (unit price <; $1) that is capable of real-time measuring bridge pier scour with resolution of ca. 2.5 cm, and overall cost for single sensor node of our work is at least 40% less expensive than existing work. The hall-effect sensor is evaluated under controlled conditions in a laboratory flume. After scour event, the typical voltage change of the hall-effect sensor is ~ 300 mV, and the system achieves signal-to-noise ratio performance of 60 dB. Finally, we also provide an equation to predict the time variation of scour depth around pier model. Moreover, our system adopts master-slave architecture has scalability and flexibility for mass deployment. We believe that our system has the potential for further widespread implementation in the field.
We design and implement an Android application, Sensorem, for efficient retrieval and visualization of wireless sensor network (WSN) data. In light of data distribution and visualization being important developmental keystones of smart cities, we seek to enhance sensor data accessibility by developing a user-friendly mobile application (Sensorem) for meaningful visualization of sensor data, targeted at maintenance personnel. Sensor selection can be made with reference to geographical location through an embedded Google Map fragment, as well as a sensor list with sensors in order of WSN node ID. Sensor data is presented through interactive graphs, and graph overlaying functions are offered for easy comparison of data trends. We also employ a Bayesian prefetch algorithm and caching mechanisms to minimize sensor data access latency, such that the app system is able to cope with network and back-end bottlenecks.
As wireless sensor network (WSN) technologies become mature, an increasing number of large-scale WSN-based long-term monitoring systems are deployed. However, data quality, especially sensor drift, is affecting the trustworthiness of sensor data. In this paper, we proposed an online algorithm to blindly calibrate sensor drift using signal space projection and Kalman filter. By utilizing data correlation among sensors, the proposed method neither requires sensors to be densely deployed nor needs prior knowledge of data models. Simulation results showed the proposed method can detect and calibrate sensor drift successfully. The mean square error of estimated drift is less than 1%, which is more accurate than existing prediction-based methods. The proposed method is also robust to measurement noise, multiplicative drift, and signal subspace estimation error.
Short-term forecasting of electric loads is an essential function required by Smart Grids. Today increasing amount of smart metering data is available enabling the development of enhanced data-driven models for short-term load forecasting. Until now, a plethora of models have been developed ranging from simple linear regression models to more advanced models such as (artificial) neural networks (NNs) and support vector machines (SVMs). Despite the relatively high accuracy obtained, the acceptance of purely data-driven models such as NN models is still remained limited due to their complexity and nontransparent nature. Therefore it is important to develop optimization schemes, which can be used to facilitate the selection of appropriate model structure resulting good forecasting accuracy with low complexity. This study presents an optimization scheme based on multi-objective genetic algorithm (GA) for designing data-driven models for short-term forecasting of electric loads. The optimization scheme is demonstrated for designing the conventional NN/MLP model using real smart metering data and weather measurements. The optimal NN model structures are identified and analyzed in terms of model complexity and forecasting accuracy.
In this paper, we present a Bayesian approach to accurately track multiple objects based on Received Signal Strength (RSS) measurements. This work shows that taking into account the spatial correlations of the observations caused by the random shadowing effect can induce significant tracking performance improvements, especially in very noisy scenarios. Additionally, the superiority of the proposed Sequential Markov Chain Monte Carlo (SMCMC) method over the more common Sequential Importance Resampling (SIR) technique is empirically demonstrated through numerical simulations in which multiple targets have to be tracked.
Many Wireless Sensor Network systems are deployed under extreme conditions. Thus, it's important to maximize their lifespan by optimal use of network's resources. Battery lifespan of a node is a crucial resource that needs to be used carefully. Energy aware routing & scheduled sensing have been introduced for careful use of battery. It is necessary to optimize battery usage by predicting it's future behavior. This will lead the users to take early decisions, thus minimizing network downtime. Hence, we explore the possibility of using meta-data on each node, to represent and predict node behavior using machine learning models. In this research, we use node voltage level as an indicator of the energy used, as voltage is proportional to available energy. Node energy consumption is modeled and predicted by ARIMA models using these voltage readings. We also classify nodes as high, medium & low use, with respect to its current and future usage, thus allowing user to take early decisions maximizing network throughput (lifetime). After evaluating predicted node behavior against a created base set of behavioral classifications, we achieved a 80% accuracy rate. Using different and increasing window sizes, we evaluated the validity of our model. In these experiments, our prediction method produced highly accurate results for all considered prediction windows. By being able to predict a node's energy consumption behavior at a higher accuracy rate, WSN users can make optimization decisions beforehand to increase the network lifetime.
Video surveillance systems are commonly used for environmental monitoring. Nevertheless, existing systems such as [1, 2] can only monitor scenes statically. When an accident or a fire event happens, we cannot get the critical pictures and check those situations immediately. Therefore, we propose to leverage pan-tilt-zoom (PTZ) cameras and develop a real-time surveillance system, where the PTZ cameras can rotate in a horizontal plane by panning and in a vertical plane by tilting, and adjust the focal length by zooming, so we can control these cameras dynamically for the emergent area. Especially for the accessibility, we implement this system as a web service to help users easily define the emergent area and preview the video of cameras remotely.
Radio Frequency Identification (RFID) is widely used in indoor positioning systems for object tracking and localization. However, there are several challenges that are yet to be addressed especially in 3-D space. When objects are cluttered densely, an arrangement synonymous to that of heaps of haphazardly arranged files in an office, locating and retrieving a file manually from the heap becomes a laborious task. In this paper, we address the challenge of localization and retrieval of objects in cluttered environments using passive RFID tags, by developing a novel indoor path loss translational model that considers the signal properties across the clutter. The proposed InPLaCE RFID system estimates the position of the object within a clutter by employing a robust translation model that accounts for the properties of the clutter and helps compensate for estimation errors over existing path loss models. Our experiments over different cluttered environments show that the proposed translational model improves the localization accuracy of objects over existing path loss models.
Outlier detection is an important task in data mining, with applications ranging from intrusion detection to human gait analysis. With the growing need to analyze high speed data streams, the task of outlier detection becomes even more challenging as traditional outlier detection techniques can no longer assume that all the data can be stored for processing. While researchers mostly focus on detecting global outliers for data streams, detecting local outliers on streaming data has been neglected. This is an example of the utility problem in machine learning, where the machine learning algorithm needs to consider how the scarcity of a critical resource in the deployment environment affects the utility of any learned model. In this paper we focus on local outliers and propose an incremental solution assuming finite memory available. Our experimental results on a variety of data sets show that our solution is well suited to application environments with limited memory (e.g., wireless sensor networks) where the state of the system is changing.
Tracking target user's indoor location with sub meter accuracy using low cost or no infrastructure is an active research topic. Indoor localization uses short range signals like WiFi, radio, ultrasound or Bluetooth signals that are affected by multipath errors and human presence. Recent studies have used stable magnetic field maps along with inertial sensors for indoor localization. In this paper, we propose a method that uses inertial sensors, magnetic field maps and indoor maps in a particle filter based implementation to improve accuracy of localization and tracking. Our method uses gradient descent algorithm to correct inaccurate user heading direction estimates due to magnetic perturbations. The tracking performance of our method is tested for four different implementations of Particle filter algorithm using magnetic maps (magnitude or vector) with and without indoor maps. We have developed an Android application to implement these four approaches and performed several experiments in a test area. The best approach out of these four achieves a mean localization accuracy of 0.75 m with a standard deviation of 0.52 m.
Wireless sensor networks usually pay attention to energy efficiency by reducing the data transmission rate or using data aggregation techniques. However, in an urgent situation of a disaster monitoring application, sensing data become very important. The sensing data are required to be at the base station quickly so that the administrators can evaluate the situation and issue the early warning message. This paper focuses on the adaptive operations of sensor nodes. The sensor nodes can automatically adjust its transmission range based on the degree of importance of sensing data. In the urgent situation, the transmission range is increased to reduce the number of hops to reduce the latency. In the normal situation, the transmission range is decreased to reduce the battery consumption of sensor nodes. The proposed mechanism, called A-TRED, evaluates the degree of importance of sensing data and adjusts the transmission range of sensor nodes accordingly. A-TRED consists of three processes which are 1) data checking process, 2) packet formation process, and 3) packet forwarding process. The simulation results show that A-TRED can automatically adapt WSN key performance factors (e.g., latency, battery consumption) by adjusting the transmission range against the environmental changes.
As cyber-physical systems (CPS) build a foundation for visions such as the Internet of Things (IoT) or Ambient Assisted Living (AAL), their communication security is crucial so they cannot be abused for invading our privacy and endangering our safety. In the past years many communication technologies have been introduced for critically resource-constrained devices such as simple sensors and actuators as found in CPS. However, many do not consider security at all or in a way that is not suitable for CPS. Also, the proposed solutions are not interoperable although this is considered a key factor for market acceptance. Instead of proposing yet another security scheme, we looked for an existing, time-proven solution that is widely accepted in a closely related domain as an interoperable security framework for resource-constrained devices. The candidate of our choice is the Web Services Security specification suite. We analysed its core concepts and isolated the parts suitable and necessary for embedded systems. In this paper we describe the methodology we developed and applied to derive the Devices Profile for Web Services Security (DPWSec). We discuss our findings by presenting the resulting architecture for message level security, authentication and authorization and the profile we developed as a subset of the original specifications. We demonstrate the feasibility of our results by discussing the proof-of-concept implementation of the developed profile and the security architecture.
A smart grid is a power system that uses information and communication technology to operate, monitor, and control data flows between the power generating source and the end user. It aims at high efficiency, reliability, and sustainability of the electricity supply process that is provided by the utility centre and is distributed from generation stations to clients. To this end, energy-efficient multicast communication is an important requirement to serve a group of residents in a neighbourhood. However, the multicast routing introduces new challenges in terms of secure operation of the smart grid and user privacy. In this paper, after having analysed the security threats for multicast-enabled smart grids, we propose a novel multicast routing protocol that is both sufficiently secure and energy efficient.We also evaluate the performance of the proposed protocol by means of computer simulations, in terms of its energy-efficient operation.
Gossip algorithms have already been identified as possible solutions for information aggregation in large-scale distributed systems. For example, the “classical” algorithm allows computation of the average of the values stored in a network, in a fully distributed manner. Smart extensions use this basic algorithm as a primitive for computing complex statistics of the values in the network, perform online optimization, system identification, etc. In this paper we bring into discussion a less used variant of gossip (SFC) which employs order statistics on exponential random variables to compute the sum of the values in a network. We study the two algorithms in parallel as they are in fact interchangeable. We show that for large-scale networks, SFC proves to be a surprisingly fast and robust aggregation method, easily extendable to allow self-stabilization. As the second contribution of the paper, we introduce such a mechanism and analyze its performance.
Most of the currently deployed integrated home management products require an experienced technician to install and configure the system. In this paper, we build upon the Internet of Things (IoT) paradigm, with the aim of delivering networked solutions that enable multi-node wireless sensor networks (WSNs) to connect to the Internet in a secure, simple and efficient way. We also describe the design and implementation of a smart-home management system. The system is composed of a lightweight tool with an intuitive user interface for commissioning of IP-enabled WSNs. The solution includes a visual programming interface with a common framework for discovering smart home services on the WSN, and a code analysis and translation engine to generate Python code. This engine analyses the application rules defined with the graphical user interface and translates them into distributed application scripts. The system also includes modules to plan the optimization of the deployment, and deploy and start the generated code. In this paper we present a prototype of the system, with the visual programming solution and code generation module.
In this paper we present a low pressure sensor based on the Pirani principle. Conversely to conventional implementations, this novel concept employs a fully wireless approach for energy supply and readout of the sensor. The energy for the used heating element is harvested from conventional power supply lines. Furthermore, we used a fully passive sensor to measure the pressure value. Both parts together offer a flexible sensor concept which can solve current implementation drawbacks of size and wired connection to this kind of sensors. The presented sensor concept is completely autonomous and offers a cheap and flexible alternative to conventional sensors even on places not easy to access and without the need for maintenance. For the whole sensor system a simulation model is derived and the model is compared to measurements. The accuracy of the model for the autonomous sensor is shown over a large area of pressure.
