J. Schoeman

Performance investigation of a sparse data compression technique with AWGN channel effects

The paper introduces a performance measure for data sources that may be used to evaluate the potential for good compression using a sparse source coding technique, as well as BER bounds in AWGN. The authors present comparative results of sparse source coding and, in terms of the introduced performance measure, complexity and error propagation in AWGN conditions. Simulation results show that the performance measure can be used effectively to predict the potential of a source for compression. Comparative PDF and BER results are shown for various sparseness levels in coded and uncoded systems

Performance evaluation of a joint source/channel coding scheme for DS/SSMA systems utilizing complex spreading sequences in multipath fading channel conditions

In recent years, the joint decoding of combined source and channel coding have shown the capability of delivering performances exceeding that of separately decoded schemes. In this paper, a well-known joint source/channel coding scheme is investigated for possible use in future 4th Generation (4G) direct sequence spread spectrum multiple access (DS/SSMA) code division multiple access (CDMA) communication systems. Due to the availability of potentially large sets of sequences with good correlation characteristics, the DS/SSMA CDMA system considered in this paper employs complex spreading sequences (CSS). A multiuser CSS-based DS/SSMA CDMA communication system was implemented as test platform for the joint source/channel coding scheme. Several families of CSSs, including Zadoff-Chu (ZC), double sideband constant envelope linearly interpolated root-of-unity filtered general chirp-like (DSB CE-LI-RU Filtered GCL), analytical bandlimited complex (ABC) and quadriphase (QPH) sequences are investigated on this platform. Simulated bit error rate (BER) performances are presented for uncoded, separate source and channel coded and joint source/channel coded systems under realistic multipath fading channel conditions.

Employing a microbolometer as a low pressure sensor

Uncooled microbolometers have become extremely popular as low cost thermal detectors used in FPAs for thermal imaging cameras. Most of the emphasis of researchers have gone towards the design and optimisation of device structures, materials, processes and readout electronics with this application in mind. However, microbolometers have the potential to be utilised towards the development of alternate applications. It is well known that the thermal conduction of microbolometers depend on the pressure surrounding the device, as this governs the dominating conduction method. This work investigates the possibility of employing a Ti thinfilm microbolometer as a low pressure sensor. A well known multi-physics simulation environment is utilised to simulate the microbolometer thermoelectric response over varied atmospheric pressure conditions. These simulation results are compared with a much simpler air pressure model than previous works using microbolometers, as well as experimental data, where the fabricated prototype showed a measured device TCR of about 0.085% K-1 and a sensitivity of about 0:701 – 10-9 W K-1 Pa-1.

An instrumentation amplifier based readout circuit for a dual element microbolometer infrared detector

The infrared band is widely used in many applications to solve problems stretching over very diverse fields, ranging from medical applications like inflammation detection to military, security and safety applications employing thermal imaging in low light conditions. At the heart of these optoelectrical systems lies a sensor used to detect incident infrared radiation, and in the case of this work our focus is on uncooled microbolometers as thermal detectors. Microbolometer based thermal detectors are limited in sensitivity by various parameters, including the detector layout and design, operating temperature, air pressure and biasing that causes self heating. Traditional microbolometers use the entire membrane surface for a single detector material. This work presents the design of a readout circuit amplifier where a dual detector element microbolometer is used, rather than the traditional single element. The concept to be investigated is based on the principle that both elements will be stimulated with a similar incoming IR signal and experience the same resistive change, thus creating a common mode signal. However, such a common mode signal will be rejected by a differential amplifier, thus one element is placed within a negative resistance converter to create a differential mode signal that is twice the magnitude of the comparable single mode signal of traditional detector designs. An instrumentation amplifier is used for the final stage of the readout amplifier circuit, as it allows for very high common mode rejection with proper trimming of the Wheatstone bridge to compensate for manufacturing tolerance. It was found that by implementing the above, improved sensitivity can be achieved.

The modelling of a capacitive microsensor for biosensing applications

Microsensing is a leading field in technology due to its wide application potential, not only in bio-engineering, but in other fields as well. Microsensors have potentially low-cost manufacturing processes, while a single device type can have various uses, and this consequently helps with the ever-growing need to provide better health conditions in rural parts of the world. Capacitive biosensors detect a change in permittivity (or dielectric constant) of a biological material, usually within a parallel plate capacitor structure which is often implemented with integrated electrodes of an inert metal such as gold or platinum on a microfluidic substrate typically with high dielectric constant. There exist parasitic capacitance components in these capacitive sensors, which have large influence on the capacitive measurement. Therefore, they should be considered for the development of sensitive and accurate sensing devices. An analytical model of a capacitive sensor device is discussed, which accounts for these parasitic factors. The model is validated with a laboratory device of fixed geometry, consisting of two parallel gold electrodes on an alumina (Al2O3) substrate mounted on a glass microscope slide, and with a windowed cover layer of poly-dimethyl-siloxane (PDMS). The thickness of the gold layer is 1μm and the electrode spacing is 300μm. The alumina substrate has a thickness of 200μm, and the high relative permittivity of 11.5 is expected to be a significantly contributing factor to the total device capacitance. The 155μm thick PDMS layer is also expected to contribute substantially to the total device capacitance since the relative permittivity for PDMS is 2.7. The wideband impedance analyser evaluation of the laboratory device gives a measurement result of 2pF, which coincides with the model results; while the handheld RLC meter readout of 4pF at a frequency of 10kHz is acceptable within the measurement accuracy of the instrument. This validated model will now be used for the geometric design and simulation of efficient capacitive sensors in specific biological detection applications.

Work in Progress - An Optimal UMTS/WCDMA Receiver Structure for Possibly Sparse Data Streams

This paper presents the work completed thus far in the design of an optimal next generation WCDMA receiver structure for situations where data streams exhibiting small or large amounts of sparseness are received. The receiver will employ a measure of sparseness, allowing for almost real-time (per timeframe) sparseness estimation to be used in an optimal adaptive estimation technique. Simulation results presented indicates the need for sparseness estimation, as well as performance benchmarks of the standard receiver structure to be used as a reference for comparison.

A novel approach to improve sparse data models

This paper implements the recently introduced measure of sparseness with a specific feedback mechanism to improve the large standard deviation that current random data sources suffer from. The authors will show the drawbacks of the current, frequently used random data sources and suggest a novel approach to rectify these disadvantages. Simulation results are presented for various blocklengths typically used in 3G UMTS/WCDMA systems, showing how the standard deviation may be significantly reduced by up to 33% by using an optimal feedback length. These optimal feedback lengths are also presented, as well as some conclusions regarding communication system simulation and future

Prevalence of canine Babesia and Ehrlichia co-infection and the predictive value of haematology

Canine babesiosis and ehrlichiosis are important tick-borne infections in South Africa. Many South African general veterinary practitioners perceive co-infection with Ehrlichia spp. as a common occurrence in dogs with babesiosis. Studies about the prevalence of co-infection in South African dogs are lacking. This retrospective study aimed to determine the prevalence of Ehrlichia co-infection in dogs with babesiosis. Additionally, the predicative value of specific haematological variables for co-infection was evaluated. The study population consisted of 205 dogs diagnosed with canine babesiosis presented to the Onderstepoort Veterinary Academic Hospital (OVAH) in 2006 and between 2011 and 2013. The Babesia-infected dogs were grouped based on presence or absence of an Ehrlichia spp. co-infection. Ehrlichia spp. co-infection was confirmed using polymerase chain reaction. Positive and negative predictive values (PPVs and NPVs) of leukopenia or thrombocytopenia for co-infection were also calculated. The prevalence of Babesia spp. and Ehrlichia spp. co-infection in this cohort of dogs was 2%. In the babesiosis dogs, the PPV of leukopenia for co-infection with Ehrlichia spp. was 1.3%, and the NPV 97.4%. Similarly, the PPV and NPVs of thrombocytopenia for co-infection were 2.1% and 100%, respectively. Co-infection with Ehrlichia spp. was a rare occurrence in dogs with babesiosis presented to the OVAH. Normal leukocyte or platelet counts confidently ruled out the presence of concurrent ehrlichiosis in this cohort of dogs. However, the diagnosis of Ehrlichia co-infection based on the presence of thrombocytopenia or leukopenia would have been associated with false positive results in more than 97.4% of cases.

Recent advances in the modelling of the thermal conductance of uncooled microbolometers

This paper presents an extensive overview and a comparison of the most popular models used to make an analytic prediction of the thermal conductance of uncooled microbolometers. The concept of an uncooled microbolometer has existed for many decades, along with simple methods to determine the thermal conduction both under vacuum and at atmospheric conditions. However, recent advances and the maturation of the MEMS fabrication technology have sparked a renewed research interest in these devices, including a number of improvements in the analytic modelling of the thermoelectric interaction, notably also improvements in the prediction of the thermal conduction. A comparison is made between a number of recent techniques, adapted for either thin-film metal or semiconductor type detector materials, operating either at vacuum conditions or atmospheric pressure conditions, as well as adjustable pressure conditions. The approach followed in the work is to determine the various parameters by means of the proposed analytic methods, which is then compared to multiphysics FEM simulation results. This comparison provides essential information regarding the shortcomings of the traditional methods. Finally, the results are also compared with experimentally extracted results from manufactured devices. The last set of results offer insight into manufacturing deviation. It is shown that the traditional methods often suffer from a significant error, sometimes in the region of almost 40%, in the prediction of the thermal conductance, which can be largely removed by using the appropriate modified analytic model.

Design issues of a low cost lock-in amplifier readout circuit for an infrared detector

In the past, high resolution thermal sensors required expensive cooling techniques making the early thermal imagers expensive to operate and cumbersome to transport, limiting them mainly to military applications. However, the introduction of uncooled microbolometers has overcome many of earlier problems and now shows great potential for commercial optoelectric applications. The structure of uncooled microbolometer sensors, especially their smaller size, makes them attractive in low cost commercial applications requiring high production numbers with relatively low performance requirements. However, the biasing requirements of these microbolometers cause these sensors to generate a substantial amount of noise on the output measurements due to self-heating. Different techniques to reduce this noise component have been attempted, such as pulsed biasing currents and the use of blind bolometers as common mode reference. These techniques proved to either limit the performance of the microbolometer or increase the cost of their implementation. The development of a low cost lock-in amplifier provides a readout technique to potentially overcome these challenges. High performance commercial lock-in amplifiers are very expensive. Using this as a readout circuit for a microbolometer will take away from the low manufacturing cost of the detector array. Thus, the purpose of this work was to develop a low cost readout circuit using the technique of phase sensitive detection and customizing this as a readout circuit for microbolometers. The hardware and software of the readout circuit was designed and tested for improvement of the signal-to-noise ratio (SNR) of the microbolometer signal. An optical modulation system was also developed in order to effectively identify the desired signal from the noise with the use of the readout circuit. A data acquisition and graphical user interface sub system was added in order to display the signal recovered by the readout circuit. The readout circuit was able to enhance the SNR of the microbolometer signal significantly. It was shown that the quality of the phase sensitive detector plays a significant role in the effectiveness of the readout circuit to improve the SNR.