Azril Haniz

Spectrum sensing on emergency radio spectrum management system

Big natural or manmade disasters damage infrastructure and cause severe effect on civilizations. Existing telecommunication infrastructure goes down as well as power, water etc. Quick and effective rescue operation requires a collaborative communication infrastructure for the rescue teams. Emergency spectrum usage of the rescue teams is one of the major issues in such scenario. Different teams interfere with each other for lack of collaboration and infrastructure limitations. This study designed and developed a cognitive wireless sensor network that collects and forms a database of the active emitters on the disaster scene. This paper focuses especially on a distributed spectrum sensing scheme for the emergency radios.

Localization of illegal radios utilizing cross-correlation of channel impulse response with interpolation in urban scenarios

Fingerprint-based localization is expected to be superior compared to other conventional range-based localization techniques especially in dense urban scenarios which are generally non line-of-sight (NLOS). However, when we consider localizing illegal radios whose parameters such as bandwidth and center frequency are unknown, there is a high possibility that the initial parameters used to create the fingerprint database in the training phase do not match the illegal radios parameters. In this paper, a novel fingerprint-based localization technique which utilizes interpolated channel impulse response (CIR) cross-correlations as fingerprints is proposed, which does not require a priori information of the illegal radio. Results show that the proposed technique is very promising in localizing illegal radios in urban NLOS environments.

A Novel Phase-Difference Fingerprinting Technique for Localization of Unknown Emitters

Localization of unknown emitters is crucial to tackle illegal radios or radio jammers which may cause harmful interference to other communication systems and disrupt services. Conventional fingerprint-based localization techniques that utilize received signal strength or channel information as location fingerprints may face problems when dealing with emitters whose parameters, such as center frequency, are unknown. In this paper, a novel localization algorithm utilizing the phase-difference between elements of an antenna array as location fingerprints is proposed. The training fingerprints are interpolated in the frequency and spatial domains before performing pattern matching with the unknown emitters fingerprints. Localization accuracy was evaluated through ray-tracing and Monte Carlo simulations, and results indicate that the proposed technique achieved relatively good performance compared to other localization techniques.

Application of geostatistical techniques for spatial interpolation of location fingerprints

Techniques such as triangulation and trilateration have been used along with RSSI, TDOA etc. for localization in outdoor scenarios. However in densely urban areas where many obstacles exist, multipaths are dominant resulting in the above techniques to be unreliable. Fingerprint-based localization may prove to be advantageous in these scenarios [1]. Location fingerprints should be collected from as many areas as possible, but this is not feasible due to time and space constraints. One approach to solve this problem is to perform spatial interpolation on the collected fingerprints to estimate the fingerprint at an arbitrary location. This research aims to provide insight on effect of the fingerprint collection intervals on estimation accuracy by performing Monte Carlo simulations under various urban environments and conditions. Popular spatial interpolation techniques include linear interpolation, nearest-neighbor interpolation and spline interpolation. In this research, we compare a technique called Kriging [2], which is a commonly used technique in the field of geostatistics, with several other common interpolation techniques. The basic idea behind Kriging-based spatial interpolation is to apply weights to already observed values at surrounding locations, and the weights are calculated according to spatial covariance values. It is assumed that propagation characteristics from points located close to each other have higher correlation than points located far away from each other. Therefore, the spatial covariance will be a decreasing function over distance, and Kriging will apply larger weights to observed values located close to the interpolation point. In Kriging, it is assumed that the observed values (fingerprints) can be expressed as the summation of a constant local mean and the residual. The expectation of the residual is assumed to be zero, and covariance of the residual between two locations is assumed to be stationary and only a function of distance. The whole interpolation process can be divided into several steps. The first step is to estimate the spatial covariance function from observed location fingerprints. Then second step is to fit the estimated covariance function with some basic covariance models. Finally, weights are calculated by solving a set of linear equations based on the covariance model obtained in the previous step. In this research, we propose and compare several variants of Kriging which may be able to capture the spatial correlation structure of the environment more accurately. The channel impulse response (CIR) and RSSI are utilized as location fingerprints. Ray-tracing simulation is conducted to obtain location fingerprints at various urban environment scenarios. Then the estimation error of several spatial interpolation algorithms is compared with respect to several factors such as fingerprint collection interval and robustness against multipath fading.