2 results for Akeila, Ehad

  • Positioning In Indoor Environments Based on INS and RF Sensor Fusion

    Akeila, Ehad (2011)

    Doctoral thesis
    The University of Auckland Library

    The past few years have witnessed an increasing demand for positioning applications in indoor environments. Several technologies have been employed to develop systems which can efficiently perform the positioning task in such environments. However, most of the available systems are either large and expensive or insufficiently accurate to be reliable for some of the critical applications. This thesis describes the development of an indoor positioning system which can provide portability, minimum cost and sufficient accuracy. Two of low cost sensor technologies have been utilised in this research; Inertial Navigation Systems (INS) and a positioning system based on the Bluetooth technology. The development of final system has been targeted by first optimizing the performance of each individual system using a series of proposed methods. Fusion of the measurements from the optimised systems in then performed using suitable fusion filters, such as Kalman and particle filter. Considering the INS based applications, a gravity compensation method is used for filtering the gravitational changes which corrupt the outputs of the accelerometers. A different method is then applied to automatically reset the INS errors found when obtaining the distance travelled by moving objects based on the measured accelerations. In enhancing the performance of the Bluetooth positioning system, a method has been developed to dynamically calibrate the radio frequency (RF) signal parameters to adapt for the environmental changes. In each of the developed methods, necessary verifications and testing have been done through simulations as well as using experimental setup designed for each of the sensor technologies. Final results show that the INS errors have been significantly reduced using the proposed resetting method which also extended the operational time from few seconds to several minutes. The performance of the Bluetooth based system has achieved positioning error of less than 1.5 metres using the proposed dynamic calibration method. Testing results of the fusion of the two optimised systems showed that the positioning error of the final system can be reduced to less than 1 metre when using either of the fusion filters. Furthermore, the fusion of the INS have demonstrated a positive impact in lowering the number of the Bluetooth reference nodes needed for achieving an adequate indoor positioning accuracy, hence cutting the overall cost when deploying the final system in real indoor applications.

    View record details
  • Testing and calibration of smart pebble for river bed sediment transport monitoring

    Akeila, Ehad; Salcic, Zoran; Kularatna, Nihal; Melville, Bruce; Dwivedi, Ambuj (2007)

    Conference item
    University of Waikato

    The Smart Pebble (smart particle), SP, has been developed for the past two years to monitor sediment transport in riverbeds. The implementation is based on use of small size and low cost acceleration and angular motion sensors. In this stage, the project is focused on calibrating and testing the final version of the SP as well as its packaging in a 4-cm diameter spherical package. The calibration was done in two stages; individual sensor calibration and complete system calibration. The complete SP unit was tested under linear motions generated by a shake table, and 2D rotational motions using two manually controlled servomotors. Offline digital signal conditioning was done in MATLAB. The preliminary results show that the system has relatively large amplitude error due to low sampling frequency. Experiments conducted by sampling a 1-Hz sinusoidal signal at different rates show that to keep the amplitude error of the system under 5% the sampling rate has to be at least 10 times the maximum bandwidth of the signals acquired from sensors.

    View record details