Akara Prayote

Detecting anomalies and intruders

Brittleness is a well-known problem in expert systems where a conclusion can be made, which human common sense would recognise as impossible e.g. that a male is pregnant. We have extended previous work on prudent expert systems to enable an expert system to recognise when a case is outside its range of experience. We have also used the same technique to detect new patterns of network traffic, suggesting a possible attack. In essence we use Ripple Down Rules to partition a domain, and add new partitions as new situations are identified. Within each supposedly homogeneous partition we use fairly simple statistical techniques to identify anomalous data. The special feature of these statistics is that they are reasonably robust with small amounts of data. This critical situation occurs whenever a new partition is added.

Two-stage spectrum sensing for cognitive radio under noise uncertainty

The robustness to uncertainty of noise power is one of main challenges to spectrum sensing technique. Since the occurrence of noise power uncertainty causes the detection performance of spectrum sensing techniques significantly degrade. In this paper, we propose two novel schemes of two-stage spectrum sensing for cognitive radio under environment as noise power uncertainty. The two-stage spectrum sensing technique combines two conventional spectrum sensing techniques to perform spectrum sensing by exploiting their individual advantages. The proposed two-stage spectrum sensing scheme exploits the merits of ED, MME and CAV techniques to determine the existence of the primary user. The ED performs spectrum sensing within a short time and offers a reliable detection at high SNRs condition. MME and CAV are robust to noise power uncertainty. Due to the combination of these techniques, the proposed schemes offer much more reliable detection when the uncertainty of noise power occurs. Even though the proposed technique takes the longest time in sensing period among two-stage spectrum sensing techniques, it is worth using this period of time to protect the primary user from harmful interference caused by the secondary user.

Network selection delay comparison of network selection techniques for safety applications on VANET

One of the most challenge applications in VANET is a safety application. The important characteristic of this application is stringent time constraints. To satisfy these constraints, seamless handoffs must be provided. The mechanism making successful seamless handoffs is a network selection technique. Several techniques such as AHP, SAW, GRA and TOPSIS were proposed to select a network satisfying application QoS and user preferences. However, these techniques have not been empirically validated and feasibly studied on embedded boxes in vehicles. Since embedded boxes have limitations on memory usages, CPU speed and etc, then the selection techniques should consume small memory usages and low computation time. In this work, the decision tree for AHP is considered since it utilizes low computation time. Moreover, the empirical comparison and feasible study between AHP, SAW, GRA, TOPSIS and the decision tree for AHP are carried out. The results show that the computation delay of decision tree for AHP is much lower than the one of pre-existing techniques. However, the proposed technuique uses little higher memory space than the others.

Fast and efficient detection of buried object for GPR image

The detection and identification of buried objects in GPR images generally involve curve fitting or pattern recognition techniques, which require high computational power and long processing time, e.g., 42 minutes per image. So far, real time detection, which requires fast processing speed, has never been achieved. This paper presents a fast and efficient technique for the detection of buried landmines in the Southern Region of Thailand. The experiment set up consists of five types of soil in the areas with varying numbers and positions of landmines. The detection ratio is high with processing time in the range of 44-240 seconds. To the best of our knowledge, this can be considered as one of the fastest detection technique.

Rice Phenology Monitoring Using PIA Time Series MODIS Imagery

This paper presents the method to determine rice cropping pattern in Thailand for future prediction of water supply demand, pricing, and other related issues including governmental policies. Datasets was obtained from an orbital instrument called a Moderate-Resolution Imaging Spectroradiometer (MODIS) operated by NASA. A Normalized Difference Vegetation Index (NDVI) was derived from MODIS datasets once every 16 days. This image data has been analyzed using image processing techniques in order to determine rice cropping area in Thailand. Rice cropping data is represented as a time series displaying type of rice crop in which peak data points indicate rice cropping cycle in each year. A Progressive Iterative Approximation (PIA) is used for signal smoothing and reducing noise by providing a Bezier curve representation of time-series data. The experimental results show that using PIA technique for noise reduction yields better results comparing with a common filtering method like Savitzky Golay filter.

Storm Intensity Estimation Using Symbolic Aggregate Approximation and Artificial Neural Network

A storm disaster is one of the most destructive natural hazards on earth and the main cause of death or injury to humans as well as damage or loss of valuable goods or properties, such as buildings, communication systems, agricultural land and etc. Storm intensity estimation is also important in evaluating the storm track prediction and risk area that will be affected by the storm. In this paper, proposed the storm intensity estimation model by using only 8 features to categorize major type of storm with symbolic aggregate approximation (SAX) and artificial neural network (ANN). The performance of the model is satisfactory, giving an average F-measure of 0.93 or 93%.

Double constraints adaptive energy detection for spectrum sensing in cognitive radio networks

In this paper, we propose double constraints adaptive energy detection (DCAED) for spectrum sensing. DCAED adapts the threshold by exploiting the interdependent between Pfa and Pd. DCAED overcomes a demerit of ED and AED in tradeoff between Pfa and Pd. The simulation results show that DCAED gives the highest detection rate, therefore the PU has the maximum protection from harmful interference. The DCAED consumes short sensing time and gives the low of false alarm rate, the SU accordingly has more opportunities to use the channel and also maintain the throughput nearly at the same rate as the other ED techniques.

Tropical cyclone track and intensity forecasting using remotely-sensed images

A tropical cyclone disaster is one of the most destructive natural hazards on earth and the main cause of death or injury to humans as well as damage or loss of valuable goods or properties, such as buildings, communication systems, agricultural land, etc. To mitigate severe impacts, track and intensity forecasting is a world-widely adopted process. With accurate forecasting, proactive measures can be appropriately applied on time to reduce both human and property losses. However, Thailand has insufficient meteorological data to apply the NWP model. In fact, the forecasting is done manually in Thailand. This makes the forecasting unreliable and time consuming, which leaves not enough time to prepare a good warning bulletin. To address these problems, this paper proposes an integrated short-range tropical cyclone track and intensity forecasting system by using only 11 features which were extracted from satellite images with improvement of the traditional statistical methods. The performance of the model is satisfactory, giving an average of 4.12 degrees of 6 hours, 12 hours and 24 hours track forecasting errors from best track data and the average errors is lower than traditional techniques by 14.16% on Mercator projection map and the average intensity forecasting errors of 6 hours, 12 hours and 24 hours lower than traditional techniques by 25.18%.

Multi-slot double constraints adaptive energy detection for spectrum sensing in Cognitive radio networks

In this paper, we propose a new scheme of an adaptive energy detection, multi-slot double constraints adaptive energy detection (MDCAED), with an objective to improve the detection performance of our previous work, double constraints adaptive energy detection (DCAED). MDCAED exploits multiple mini-slot technique, which achieves the diversity reception concept, to increase the ability to distinguish noise from the PU since the effect of diversity reception increases the received SNR. MDCAED performs spectrum sensing by spitting a sensing slot into a multiple mini-slot. Each mini-slot is performed spectrum sensing using DCAED and the final decision is made by using K-of-N rule. The decision threshold is adapted on the SNR of each mini-slot using DCAED. Therefore, by exploiting the multiple mini-slot concept together with DCAED, Pfa is reduced while Pd is improved. Although, the detection performance is improved, the average sensing time of MDCAED slightly increases compared to DCAED since the system threshold needs to adapt more than once. Nevertheless, the average sensing time of MDCAED still achieves the spectrum sensing requirement.

Application of remote sensing for tropical cyclone track forecasting based on statistical methods

A tropical cyclone disaster is one of the most destructive natural hazards on earth and the main cause of death or injuries to humans as well as damages or losses of valuable goods or properties, such as buildings, communication systems, agricultural land, etc. To mitigate severe impacts, the accuracy of track forecasting model is world-widely developed and improved. The accuracy of tropical cyclone track forecasting is very important for risk area evaluation that will be affected by the tropical cyclone due to evacuation in time can reduce both human and property losses. However, Thailand has insufficient meteorological data to apply the numerical weather prediction models. In fact, the forecasting in Thailand is done manually. This makes the forecasting unreliable and time consuming, which leaves not enough time to prepare a warning bulletin. To address these problems, this paper proposes an integrated short-range tropical cyclone track forecasting system which analyzes tropical cyclone tracks from available satellite images. The performance of the model is satisfactory, giving an average of 4.92 degrees of 6 hours, 12 hours, 24 hours, 48 hours and 72 hours forecasting errors from best track data and the average error is lower than traditional techniques by 25.45% on Mercator projection map.