Search

The next stage performance evaluation of an athlete can be predicted by implementing Artificial Intelligence technique. In track cycling event, coach and sports physician are concerned with the per...

Gas distribution mapping (GDM) learns models of the spatial distribution of gas concentrations across 2D/3D environments, among others, for the purpose of localizing gas sources. GDM requires run-t ...

Closed circuit television cameras (CCTV) are widely used in monitoring. Most rely on human operators and controllers for detection and information gathering, e.g. the whether there is a person in the monitored area and how many people may be present. This paper presents an adaptive crowd counting system based on two algorithms that estimate the density of each pixel in each frame and use it as a basis for counting people. One algorithm uses scale-invariant feature transform (SIFT) features and clustering to represent pixels of frames (SIFT algorithm) and the other uses features from accelerated segment test (FAST) corner points with SIFT features (SIFT-FAST algorithm). Each algorithm is designed using a novel combination of pixel-wise, motion-region, grid map, background segmentation using Gaussian mixture model (GMM) and edge detection based on Canny algorithm. The Mall and University of California (UCSD) datasets have been used to evaluate the proposed algorithms. Results show that their average accuracies and processing times for each dataset are similar but their accuracies at frame level are different. A fusion technique is proposed and used to increase accuracy by combining the result of the algorithms at frame level. The mean deviation error and the mean absolute error for the two proposed algorithms are less than 0.1 and 3.1, respectively, for the Mall dataset and less than 0.07 and 1.9, respectively, for UCSD dataset.

Assessing the smell of burning is vital, as it can help to further detect and prevent early fire. In this paper, an early stage fire detection algorithm has been introduced using Fuzzy k- Nearest Neighbour (FkNN). The tests were made on normally available seven fire sources and three building structure resources. All the test samples were scorched in a vacuum oven at various temperature points, pushed out using vacuum pumps to be sniffed by the electronic nose. The experiments were done in a confined room with monitored temperature and humidity level. Time domain data were sampled. Prior to be given to the classifier, the smellprints were normalised and the features were extracted. Experimental classification results show that the integration of fuzzy logic into conventional kNN has enhanced the accuracy of the classifier and gave excellent consistency, nonetheless of humidity and temperature disparity, baseline sensor errors, the diverse emission concentration range and varied scorching temperature levels. The average classification precision for the classification system is 96.15%.

Identification of burning smell is crucial because enables early fire recognition and avoidance. Based on this study, an early stage fire detection algorithm is offered via Probabilistic Neural Network (PNN). Experiments were conducted on seven generally accessible flammable materials and three building construction materials. All the materials were scorched in a vacuum oven at various temperature points, pushed out using vacuum pumps to be sniffed by the electronic nose. The experiments were done in a confined room with monitored temperature and humidity level. Standardised feature extractions of the smell print data were carried out prior to subjection of detection categorization. These aspects signify the odour pointers within the time frame. Experimental categorization outcomes indicate that the tuning of spread factor in PNN classifier has enhanced the precision of classification and delivered excellent reliability, irrespective of humidity variation and ambient temperature, the various gas concentration levels, exposure towards different heating temperature range and baseline sensor drift. The mean classification accuracy for this classification model has been identified at 94.18%.

This chapter overviews the issue of multicollinearity in electronic nose (e-nose) classification and investigates some analytical solutions to deal with the problem. Multicollinearity effect may harm classification analysis from producing good parameters estimate during the construction of the classification rule. The common approach to deal with multicollinearity is feature extraction. However, the criterion used in extracting the raw features based on variances may not be appropriate for the ultimate goal of classification accuracy. Alternatively, feature selection method would be advisable as it chooses only valuable features. Two distance-based criteria in determining the right features for classification purposes, Wilk's Lambda and bounded Mahalanobis distance, are applied. Classification with features determined by bounded Mahalanobis distance statistically performs better than Wilk's Lambda. This chapter suggests that classification of e-nose with feature selection is a good choice to limit the cost of experiments and maintain good classification performance.

The lack of information on ground truth gas dispersion and experiment verification information has impeded the development of mobile olfaction systems, especially for real-world conditions. In this paper, an integrated testbed for mobile gas sensing experiments is presented. The integrated 3 m × 6 m testbed was built to provide real-time ground truth information for mobile olfaction system development. The testbed consists of a 72-gas-sensor array, namely Large Gas Sensor Array (LGSA), a localization system based on cameras and a wireless communication backbone for robot communication and integration into the testbed system. Furthermore, the data collected from the testbed may be streamed into a simulation environment to expedite development. Calibration results using ethanol have shown that using a large number of gas sensor in the LGSA is feasible and can produce coherent signals when exposed to the same concentrations. The results have shown that the testbed was able to capture the time varying characteristics and the variability of gas plume in a 2 h experiment thus providing time dependent ground truth concentration maps. The authors have demonstrated the ability of the mobile olfaction testbed to monitor, verify and thus, provide insight to gas distribution mapping experiment.

In this study, an early fire detection algorithm has been proposed based on low cost array sensing system, utilizing gas sensors, dust particles and ambient sensors such as temperature and humidity sensor. The odor or smell-print emanated from various fire sources and building construction materials at early stage are measured. For this purpose, odor profile data from five common fire sources and three common building construction materials were used to develop the classification model. Normalized feature extractions of the smell print data were performed before subjected to prediction classifier. These features represent the odor signals in the time domain. The obtained features undergo the proposed multi-stage feature selection technique and lastly, further reduced by Principal Component Analysis (PCA), a dimension reduction technique. The hybrid PCA-PNN based approach has been applied on different datasets from in-house developed system and the portable electronic nose unit. Experimental classification results show that the dimension reduction process performed by PCA has improved the classification accuracy and provided high reliability, regardless of ambient temperature and humidity variation, baseline sensor drift, the different gas concentration level and exposure towards different heating temperature range., Comment: electronic nose; gas sensors; fire detection; feature selection; feature fusion; Artificial intelligence, machine learning, neural networks, remote sensing, decision support

This paper presents a performance analysis of two open-source, laser scanner-based Simultaneous Localization and Mapping (SLAM) techniques (i.e., Gmapping and Hector SLAM) using a Microsoft Kinect to replace the laser sensor. Furthermore, the paper proposes a new system integration approach whereby a Linux virtual machine is used to run the open source SLAM algorithms. The experiments were conducted in two different environments, a small room with no features and a typical office corridor with desks and chairs. Using the data logged from real-time experiments, each SLAM technique was simulated and tested with different parameter settings. The results show that the system is able to achieve real time SLAM operation. The system implementation offers a simple and reliable way to compare the performance of Windows-based SLAM algorithm with the algorithms typically implemented in a Robot Operating System (ROS). The results also indicate that certain modifications to the default laser scanner-based parameters are able to improve the map accuracy. However, the limited field of view and range of Kinect’s depth sensor often causes the map to be inaccurate, especially in featureless areas, therefore the Kinect sensor is not a direct replacement for a laser scanner, but rather offers a feasible alternative for 2D SLAM tasks.

Nowadays, chili is an important and high value product that able to give higher returns to farmers. However, chili plant fruitfulness should be given priority so that it's not damaged by pets and diseases. There are a few diseases that could attack the chili plant through the leaves. This research paper describes an image processing technique that identifies the visual symptoms of chili plant diseases using an analysis of colored images. This project proposed the design of software program that recognizes the color and shape of the chili leaf image. A few problems and constraints had to be identified before starting the project such as the different color of chili leaf, shape of chili leaf taken in different angle and distance, the group of the chili leaf and the resolution of the image captured. LABVIEW software is used to capture the image of chili plant in RGB color model and MATLAB software is used to enable a recognition process to determine the chili plant disease through the leaf images. The image recognition processes include the threshold, complementation, edging, segmentation, colors comparison and colors recognition. The recognition result of this research is about 93.3% from 120 images of chili plant. The proposed method in recognizing chili plant disease is demonstrated by experiments.

This paper describes the web services designed to handle complex and compute intensive web data queries for resources constrained devices in Internet-of-Things. The compute and memory intensive task of extracting useful data from webpages are done by online cloud server hosting the service. Essential data query patterns, filtering and manipulation techniques applied to break and reduce webpage content to minimal and easy to use data format, such as Javascript Object Notation (JSON) or Comma Separated Values (CSV). Finally, an easy to use web based interface developed to assist user in sequencing and formulating the query configurations for their things. The web query service enables IoT devices to be able to extract data and respond to data fetched from almost any websites in the web. Caching mechanism implemented as well to reduce server load on frequently requested resources from multiple things.

The European Logarithmic Microprocessor (ELM) had been an outstanding breakthrough in logarithmic number system (LNS) research history. The processor successfully reaches the par ability of floating-point (FLP) processor with its rapid and accurate design towards FLP. The design was able to improve the LNS addition and subtraction procedure, which are the drawbacks of any implementation of LNS arithmetic. ELM's subtraction operation had adopted a unique approach, which is the first-order co-transformation to overcome the singularity-to-zero issue of the non-linear function in negative region. Therefore, this research had been introduced to extensively compare and analyze the ELM-based addition and subtraction procedures with the same co-transformation technique implemented in positive region. In achieving this, two aspects are considered, which are the accuracy towards FLP and the memory consumption of both procedures in both regions. Conclusively, the exact ELM-based implementation in positive region of both operations could be realized and achieved comparable accuracy and memory area with a slight modification of the operation procedure. The outcome of this analysis could benefit further investigation of optimizing the LNS design for hardware implementation.

Interpolation is among of the most popular approach used in approximating the values in logarithmic number system (LNS) arithmetic design. This method that often combines with lookup tables (LUTs) manages to produce efficient LNS design in area, latency and accuracy. Current research proves that the combination of interpolators with co-transformation in LNS subtraction had shown extreme improvements in terms of speed and area, which is comparable to floating point (FLP) performance. Taking the advantage, this research had been conducted to analyze the implementation of these three interpolators, which are Taylor, Lagrange and modified Lagrange, in a 32-bit environment of the LNS addition and subtraction procedures with the first-order co-transformation in positive region. The designs were analyzed in two categories, which are the accuracy towards FLP and the total memory consumption. The best interpolator was selected based on the most optimum area consumption design with manageable accuracy in FLP limit. The outcome of this analysis could benefit further improvements in LNS design.

The right training programs are an important factor to increase the cycling performance among the professional track cyclist. Over the years, the cyclist performance was based on the feedback from bicycle’s kinematics and physiological condition. The advancement in sensor technologies allows the optimization of the training program; by combining both information from the cyclist’s physiological condition and kinematic data from the bicycle. The physiological conditions such as heart rate variability (HRV) and forehead temperate can be combined with bicycle kinematic data such as speed and distance to provide accurate assessment of the track cyclist’s condition and training program intensity. A system that combines data from physiological signal and bicycle kinematic has been developed for this purpose. Wearable physiological body sensors and bicycle kinematic sensors are deployed using wireless sensor network (WSN). HRV provide using photoplethysmography (PPG) technique that capture signal from cyclist’s finger, which provide 3 % error rate refer to heart rate belt. Data handling and communication was developed based on Zigbee protocol whereby the WSN centralized base-station was supported by two repeater node which was used to extend signal coverage in Velodrome to prevent data losses. With two repeater nodes and adjustment on the routing protocol, the packet drops were reduced from 46 to 3 %. The propagation study was carried out in the Velodrome with environment temperature range from 28 to 30 °C and humidity was observed at 85 %. The optimization of network topology by considering the connectivity among the wireless nodes is crucial in order to reduce data losses.

Linear discriminant analysis (LDA) has been widely used in the classification of multi sensor data fusion. This paper discusses the performance of LDA when the classifications were performed based on feature extraction and feature selection methods. Comparisons were also made based on single sensor modality. These strategies were studied using a honey dataset along with two types of sugar concentration collected from two types of sensors namely electronic nose (e-nose) and electronic tongue (e-tongue). Assessment of error rate was achieved using the leave-one-out procedure.

Microprocessors have grown tremendously in its computing and data crunching capability since the early days of the invention of a microprocessor. Today, most microprocessors in the market are at 32 bits, while the latest microprocessors from IBM, Intel and AMD are at 64 bits. To further grow the computational capability of a microprocessor, there are two possible paths. One method is to increase the data bus size of the microprocessor to 128/256/512 bits. The larger the data bus size, the more data can be crunched at any one time. The second method is to implement multiple microprocessor core in a single microprocessor unit. For example, Intel's Pentium 4 Dual Core and AMD's Athlon Dual Core both have two microprocessor core within a single microprocessor unit. Latest from Intel and AMD are quad core microprocessors with four microprocessor core within a single microprocessor unit. Both methods have its advantages and disadvantages. Both methods yields different design issues and have different engineering limitations. This research looks into the possibility of implementing a large data bus size VLIW microprocessor core of 256 bits on the data bus. VLIW is chosen as opposed to CISC and RISC due to its ease of scalability.

Logarithmic number system or LNS has become an optimal choice in digital image processing instead of floating point (FP) system based on latest researches in LNS. Digital image processing which deals with a lot of complex operations such as multiplication and division, makes LNS as a great choice of implementation. However, the implementation had been restricted by the addition and subtraction function in LNS arithmetic as these functions entail complex procedures and circuitry. As its huge potential to be a substitution of FP, there is an urgent need for LNS to improve the performance of both operations. Hence, various studies had been conducted in this area, however most of the research concern the implementation of these operations in the negative region. Therefore, this study is conducted with the objective on the exploration of both LNS addition and subtraction operations in the positive region and highlights the potential areas for design modifications and improvements. Then, these enhancements will be combined with other arithmetic functions in creating an optimum LNS design to be utilized in any digital image processing system.

The ubiquitous increase in wireless technologies utilizing the 2.4 GHz frequency band leads to spectrum congestion that reduce Quality of Service (QoS) of wireless technologies especially IEEE802.15.4 Wireless Sensor Network (WSN) due to its low data rate and low transmission power characteristics. A number of research proposed channel separation by utilizing Energy Detection (ED) as a solution to overcome the interference issue. However, to the author's knowledge there is little information on scan duration set for ED that need to be pre-configured before any communication or transmission can be established. This paper present an empirical study conducted to identify suitable scan duration for ED. Experimental test in this study is conducted in two different conditions: 1) without interference and 2) with interference from IEEE802.11b WiFi Access Point (AP). After considering the tradeoff between accuracy and false detection, the outcome from this study suggests that n=3 is suitable scan duration (SD) to be configured for ED.

Multiple weather stations and Wireless Sensor Network (WSN) system deployed throughout the state of Perlis to monitor weather and environmental conditions. Weather data from sensor nodes hops through WSN nodes towards base station attached to embedded computer system. These data accessible by eKoView, web-based monitoring application provided. However, current system limits monitoring to local area connected by WSN nodes. Since, weather stations operate at multiple geographical areas beyond nodes range, an online cloud server bridge these gaps to collect sensor data from local base stations to centralized database. Online users can access website with WSN base stations marked on Google Map and dynamically plot graph of historical weather data.

Almost all robotics applications require accurate robot positioning. However, most of the developed methods lacks in ground truth reference to verify its accuracy relative to the real world. This paper proposes an effective vision based system to accurately track mobile robot's true position and orientation using multiple overhead cameras. This system is able to track and localize multiple mobile robots simultaneously within a 3m × 6m arena. Images from the cameras are calibrated using calibration grid image to remove fish eye effect and further calibrated based on point coordinates (x, y) to eliminate camera angle distortion error. Each robot is assigned with a symbol marker for identification. A geometric feature based pattern matching algorithm is used to track the markers position and orientation. Data obtained from all four cameras are merged according to its relative offsets to obtain localization in a global coordinate frame. The developed system is able to localize multiple robots with errors of less than 1 cm and 1°.

This paper presents the study of video streaming over wireless channel based on experimental measurements in the presence of fading caused by the physical environmental. The emulation of video streaming file through wireless channel is measured using IxChariot software from Ixia. The obtained emulation of signal quality from the video streaming file was measured in terms of network throughput, RSSI and packet loss. The results show the credibility of wireless network for streaming video file in agricultural area. The statistical results show that there is significant negative effect of physical environmental condition on wireless video streaming and received video quality.

The feasibility of using Kinect sensor for 2D Simultaneous Localization and Mapping (SLAM) application has been widely studied. Researchers concluded that the acquired maps are often inaccurate due to the limited field of view of the sensor. Therefore in this work, we complemented the Kinect with a laser scanner and proposed a method to merge the data from both sensors. Two SLAM algorithms (i.e Gmapping and Hector SLAM) were tested using the method, in different environments. The results show that the method is able to detect multi-sized objects and produce more accurate map as compared to when using single sensor (i.e Kinect only or laser scanner only). Finally, the performance of the Gmapping and Hector SLAM are compared particularly in terms of the computational complexity and the map accuracy.

This paper evaluates the performance of parametric and non-parametric classification rules in sensor technology. The growing of sensor technologies, e-nose and e-tongue, has urged engineers to equip themselves with the utmost recent and advanced statistical approaches. As data collected from e-nose and e-tongue face some complexities, often data pre-processing and transformation are performed prior to the classification. This paper discusses the comparisons made on some known parametric and non-parametric classification rules in the application for classifying data of e-nose and e-tongue. The comparisons which based on leave-one-out accuracy, sensitivity and specificity shows that non-parametric approaches especially k-nearest neighbour does not much distorted with changes of distribution, but Naive Bayes is greatly influenced by the structure of the data.

A disposable screen-printed multi channel taste sensor composed of several types of lipid as transducers and a computer as data analyzer could detect taste in a manner similar to human gustatory sensation. The disposable taste sensor was used to measure the electrical potential resulted from the interaction between lipid membranes and taste substances. In the present study, two types of packaged commercial milk, the ultra high temperature (UHT) and the pasteurized milk were tested. It was found that the disposable taste sensor is capable to discriminate reliably between fresh and spoiled milk and to follow the deterioration of the milk quality when it is stored at room temperature based on a pattern recognition principle namely Principle Component Analysis (PCA). This research could provide a new monitoring method ideally for simple and cheap decentralized testing for controlling the quality of milk, which may be of great use in the dairy industries.

Heart rate variability (HRV) extracted from human physiological turn to be important data as indicators of human health. Originally, HRV calculated and monitored from Electrocardiography (ECG) signal. This research proved that photoplethysmography (PPG) as a simple non-invasive method can measure HRV in motion condition. PPG used Light Photo Sensor (LPS) to capture light reflectance intensity on the finger. The signal is sampled, filtered, processed and sent wirelessly through ZigBee protocol. The system applied on glove to make it wearable and easy to wear. Experimental results show that the system functioning properly with HRV received, displayed and recorded by the base station.

Traditionally, the variety of animal species are accessed using conventional method that is normally costly, consumed a lot of time and the most important missing element is that it does not equipped with any permanent recording device. In order to replace this conventional method, an automated sound recording device is developed. This device has the capability to record the animal vocalization and has the power system to be used in remote area. This paper presents the work of sound analyzers for acoustic and health monitoring of the forest. The sound analyzer is important for the researcher as it can be used for further analysis such as frequency of animal vocalization and pattern recognition. The developed system uses the LabVIEW (Laboratory Virtual Instrument Engineering Workbench) platform.

Current lifestyles promote the development and advancement in wireless technologies, especially in Wireless Sensor Networks (WSN) due to its several benefits. WSN offers a low cost, low data rate, flexible routing, longer lifetime, and low-energy consumption suitable for unmanned and long term monitoring. Among huge WSN applications, some key applications are smart houses, environmental monitoring, military applications, and other monitoring applications. As a result, ubiquitous increase in the number of wireless devices occupying the 2.4GHz frequency band. This causes a dense wireless connection followed by interference problem to WSN in the 2.4GHz frequency band. WSN is most affected by the interference issue because it has a lower data rate and transmission power compared to WLAN. Despite efforts made by researchers, to the author's knowledge, the interference issue is still a major problem in wireless networks. This paper aims to review the coexistence and interference issues of existing wireless technologies in the 2.4GHz Industrial, Scientific and Medical (ISM) band.

Current studies document the effectiveness of multi-sensing technique implementation to mimic or to complement human senses. This work demonstrated the successful application of multi-sensing techniques such electronic tongue (e-tongue), electronic nose (e-nose) and Fourier Transform Infrared Spectroscopy (FTIR). The fusion of these modalities enhance the classification of pure Tualang honey using Linear Discriminant Analysis (LDA), Probabilistic Neural Network (PNN), Support Vector Machine (SVM) and k-Nearest Neighbour (KNN). KNN and PNN are able to classify between pure and adulterated honey samples, outperform LDA and SVM. By performing data fusion, SVM and LDA classifier can achieved more than 80% accuracy, while KNN and PNN obtained greater precision, up to 96% correct classification. The findings confirmed that, multi-sensing technique; either KNN or PNN was significantly superior compared to SVM and LDA classification methods. Thus, both analyses are able to discriminate between pure and adulterated honey.

Logarithmic number system (LNS) has been a trend in digital signal processing (DSP) for recent years, particularly digital image processing. LNS was been implemented in DSP processors and even as an enhancer tool in image improvements. The selection of LNS is due to the ease of operation for multiplication, division, square and square-root which been replaced by fixed-point addition, subtraction, and left- and right-shift operations, respectively. Current researches have found that LNS is possible to be a competitor and a potential replacement of floating point (FP) system in near future, as it provides comparable strength towards the latter. The other arithmetic operations, which is the addition and subtraction, however, have to face great challenges in implementations as it requires complex procedures and circuitry. As LNS potential to be a substitution of FP, there is a significant need for LNS to improve the performance. Hence, this paper outlines the evolution of the LNS architecture design and highlights the potential areas for further improvements. These enhancements will then be implemented and verified in a specific digital image processing system.

Determination of burning smell is important because it can help in early fire detection and prevention. In this paper, a household burning smell classification system for early fire detection application has been proposed using Probabilistic Neural Network (PNN) and PCA analysis. The experiments were performed on recorded smell samples from combustion of ten different commonly available household, including candle, joss sticks, air freshener, mosquito coil, newspaper, card board, plastic materials, Styrofoam and wood. All the experiments were done in a test chamber with humidity and temperature sensors. Portable Electronic Nose (PEN3) from Airsense Analytics is used as the measurement device. The smell source is placed 0.3m from the PEN3 and the time-series signal is measured for two minutes. The odour metrics is modelled using Probabilistic Neural Network. It is found that the average classification accuracy for the model is 99.62%.

The field of measurement technology in the sensors domain is rapidly changing due to the availability of statistical tools to handle many variables simultaneously. The phenomenon has led to a change in the approach of generating dataset from sensors. Nowadays, multiple sensors, or more specifically multi sensor data fusion (MSDF) are more favourable than a single sensor due to significant advantages over single source data and has better presentation of real cases. MSDF is an evolving technique related to the problem for combining data systematically from one or multiple (and possibly diverse) sensors in order to make inferences about a physical event, activity or situation. Mitchell (2007) defined MSDF as the theory, techniques, and tools which are used for combining sensor data, or data derived from sensory data into a common representational format. The definition also includes multiple measurements produced at different time instants by a single sensor as described by (Smith & Erickson, 1991).

The perceived quality of fruits, such as mangoes, is greatly dependent on many parameters such as ripeness, aroma, firmness, shape, size, and is influenced by other factors such as harvesting time. Unfortunately, a manual fruit grading has several drawbacks such as subjectivity, tediousness and inconsistency. By automating the procedure, as well as developing new classification technique, it may solve these problems. This paper presents the novel work on the bio-inspired multi-modality sensing system for classification and quality assessment of mangoes cv. Harumanis Mango using charge coupled device (CCD) camera and Infrared (IR) camera. A Fourier-based shape separation method was developed from CCD camera images to grade mango by its shape and able to correctly classify 100%. Colour intensity from infrared image was used to distinguish and classify the level of maturity and ripeness of the fruits. The finding shows 92% correct classification of maturity levels by using infrared vision.

Producing chili is a daunting task as the plant is exposed to the attacks from various micro-organisms and bacterial diseases and pests. The symptoms of the attacks are usually distinguished through the leaves, stems or fruit inspection. This paper discusses the effective way used in performing early detection of chili disease through leaf features inspection. Leaf image is captured and processed to determine the health status of each plant. Currently the chemicals are applied to the plants periodically without considering the requirement of each plant. This technique will ensure that the chemicals only applied when the plants are detected to be effected with the diseases. The image processing techniques are used to perform hundreds of chili disease images. The plant chili disease detection through leaf image and data processing techniques is very useful and inexpensive system especially for assisting farmers in monitoring the big plantation area.

Electronic Nose (e-nose) is an intelligent instrument that is able to classify different types of odours. The e-nose applications include food quality assurance, fragrance industry, medical diagnosis, environmental monitoring, agricultural industry and homeland security. The current enose design trend are portable, small size, low power consumption, high processing power using embedded controller and easy to operate to enable it to perform the designed tasks effectively. This paper deals with the design issues of a hand-held e-nose based on sensor selection and optimum embedded controller capabilities. A summary of proposed hardware and software solutions are provided with emphasis on data processing. The data processing utilizes multivariate statistical analysis i.e. Principal Component Analysis (PCA), Hierarchical Cluster Analysis (HCA) and Linear Discriminate Analysis (LDA). The developed instrument was tested to discriminate the Ganoderma boninense fruiting body (basidiocarp). Initial results show that the instrument is able to discriminate the samples based on their odour chemical fingerprint profile.

To provide reliable and adequate network coverage whilst minimizing the cost of wireless sensor network (WSN) deployments, detailed knowledge of wireless signal propagation within the specific environments is required. There are many WSN devices on the market that have been developed using proprietary systems and therefore have different performances, although implementing similar standards. This paper presents a comparative performance measurement and analysis of three types of WSN devices evaluated for application in a mixed-crop farm. The results show that the Xbee-PRO maintains very strong RSSI values in open field measurements that are sometime 15 dBm higher than those obtained from the IRIS and Microchip motes. Overall, two important factors that influence WSN node performances are antenna height and the type of antenna used. Whip omni-directional antenna has been shown to double the range of the WSN node compared to a patch antenna. Results also show that the log-distance propagation model is a more flexible model that can be used to model a variety of channels, although it lacks standard global parameter values.

We present the design of an efficient multiplication unit. This multiplier architecture is based on radix 4 booth multiplier. In order to improve his architecture, we have made 2 enhancements. The first is to modify the Wen-Chang's modified booth encoder (MBE) since it is the fastest scheme to generate a partial product. However, when implementing this MBE with the simplified sign extension (SSE) method, the multiplication's output is incorrect. The 2nd part is to improve the delay in the 4:2 compressor circuit. The redesigned 4:2 compressor reduced the delay of the carry signal. This modification has been made by rearranging the Boolean equation of the carry signal. This architecture has been designed using Quartus II. The Gajski rule has been adopted in order to estimate the delay and size of the circuit. The total transistor count for this new multiplier is being a slightly bigger. This is due to the new MBE which is uses more transistor. However in performance speed, this efficiency multiplier is quite good. The propagation delay is reduced by about 2% - 7% from other designers.

A disposable masfe sensot fabncared using screen-printing rcchnologi was developed to test vadous types of commercial milk products. The anay of sensors composed of several kinds of lipid as msdvcers and a computer IS data analyzer could detecr msce in a manner similar to human gustatory sensation. The sensing principlc of the scicen-ptinred taste sensor was based on measurement of the elecaid potential resulred from interaction between lipid membranes and taste subsmces. Coupled with pattern recognition tools namely Principle Component Analysis @CA) for non-supervised tcchniquc and Discriminant Analysis (DA) for supervised technique, it displayed the capabilities of the sensors. In the present sNdy, two types of packaged commercial milk, the ulua high tUnperaNCe (UHT) and the pasteurized milk were rested using the disposable tadsee sensor. The srudy showed that different brands of milk were easily distinguished by DA and the dassification between fresh and spoiled milk were projected using PCA. The taste sen501 was also used to follow the deterioration of rhc qyaliry of milk when it is rrorcd ac room tempcramre. ?he dam obtincd were usmd with ptincipal component analysis and, the detenontion pattern cadd clearly be faUowed in the diagrams. This rcsearch could provide a I~RV monitoring method id+ lor simplt and chhcip decentralized testing for controlling the q~9ry of milk which may be of great use in the deiv induse. Kepvordr : Disporiblc taw scnsoq Mk qdty control; Pmm rccopition

Electronic nose (e-nose) is an instrument that mimics human olfactory system and used to detect, discriminate and classify odours. The instrument applications include food quality assurance, plant disease detection and environmental monitoring. Recent developments in embedded technology have made possible to develop a cost-effective instrument with simple operating procedures. This paper describes the selection of an optimum embedded controller for the development of a hand-held e-nose. The developed instrument uses off-the-shelf components, i.e., sensors, microcontroller and signal conditioning circuit. The data processing utilises multivariate statistical analysis, i.e., principal component analysis (PCA), hierarchical cluster analysis (HCA) and linear discriminate analysis (LDA). The developed instrument was tested to discriminate the basic artificial food essences. Initial results show that the instrument is able to discriminate the samples based on their odour chemical fingerprint profile. The multivariate statistical analysis plots show that the samples are able to be grouped into different clusters.