Proceedings of the Australasian Computer Science Week Multiconference
High-dimensional data is becoming more and more available due to the advent of big data and IoT. ... more High-dimensional data is becoming more and more available due to the advent of big data and IoT. Having more dimensions makes data analysis cumbersome increasing the sparsity of data points due to the problem called “curse of dimensionality“. To address this problem, global dimensionality reduction techniques are used; however, these techniques are ineffective in revealing hidden outliers from the high-dimensional space. This is due to the behaviour of outliers being hidden in the subspace where they belong; hence, a locally relevant subspace is needed to reveal the hidden outliers. In this paper, we present a technique that identifies a locally relevant subspace and associated low-dimensional subspaces by deriving a final correlation score. To verify the effectiveness of the technique in determining the generalised locally relevant subspace, we evaluate the results with a benchmark data set. Our comparative analysis shows that the technique derived the locally relevant subspace that consists of relevant dimensions presented in benchmark data set.
High-dimensional data brings new challenges and opportunities for domains such as clinical, scien... more High-dimensional data brings new challenges and opportunities for domains such as clinical, scientific and industry data. However, the curse of dimensionality that comes with the increased dimensions causes outlier identification extremely difficult because of the scattering of data points. Furthermore, clustering in high-dimensional data is challenging due to the intervention of irrelevant dimensions where a dimension may be relevant for some clusters and irrelevant for others. To address the curse of dimensionality in outlier identification, this paper presents a novel technique that generates candidate subspaces from the high-dimensional space and refines the identification of potential outliers from each subspace using a novel iterative adaptive clustering approach. Our experimental results show that the technique is effective.
This paper presents the design of ladder-type filters based on film bulk acoustic wave resonator ... more This paper presents the design of ladder-type filters based on film bulk acoustic wave resonator (FBAR) in Ku-band. The proposed FBAR filter has an insertion loss of −3 dB, out-of-band rejection of −12 dB and 3 dB bandwidth of 1.0 GHz from 15 GHz to 16 GHz. Based on the characteristics of the FBAR filter, the expected characteristics of FBAR resonators are determined by using the 1D numerical analysis. This design proves that it is possible to design a wide-bandwidth FBAR filter in Ku-band.
Wireless biosensor networks (WBSNs) collect information about biological responses and process it... more Wireless biosensor networks (WBSNs) collect information about biological responses and process it using scattered battery-power sensor nodes. Such nodes demand ultra low-power consumption for longer operating time. Ultra Wide Band (UWB) is a potential solution for WBSNs due to its advantage in low power consumption at reasonable data rate. However, such UBW technology requires high quality (Q) factor passive components. This paper presents detailed analysis, design and optimization of physical parameters of silicon-on-sapphire (SOS) and micro-electro-mechanical-systems (MEMS) inductors for application in UWB transceivers. Results showed that the 1.5 nH SOS inductor achieved Q factor of 111 and MEMS inductor achieved Q factor of 45 at 4 GHz frequency. The voltage controlled oscillator (VCO) designed with SOS inductor achieved more than 10 dBc/Hz reduction in phase noise and consumed half the power compared to VCO with MEMS inductor. Such low power VCO will improve battery life of a U...
This paper presents architecture independent characterization of embedded Java workloads based on... more This paper presents architecture independent characterization of embedded Java workloads based on the industry standard GrinderBench benchmark which includes different classes of real world embedded Java applications. This work is based on a custom built embedded Java virtual machine (JVM) simulator specifically designed for embedded JVM modeling and embodies domain specific details such as thread scheduling, algorithms used for native
In accordance with the object-oriented design principles, the flight simulation platform of each ... more In accordance with the object-oriented design principles, the flight simulation platform of each function is modular and hierarchical. The 1 Pool 2 Pieces 4 Layers software framework is proposed, using VC++ features to abstract all the replaceable modules to the same class, realizing rapid conversion of a variety of aircraft and different tests, to achieve reconfigurable. The realistic flight dynamics model considering wind disturbance is established, and with F-16 model as an example has conducted digital simulation and visual simulation. The results show that the platform has good maneuverability and fidelity, and provides reliable basis for aircraft design.
11th IEEE International Symposium on Personal Indoor and Mobile Radio Communications. PIMRC 2000. Proceedings (Cat. No.00TH8525)
This paper describes the design and performance of a mixed signal, very high speed, analog-to-dig... more This paper describes the design and performance of a mixed signal, very high speed, analog-to-digital converter implemented using gallium arsenide MESFET technology for software radio applications. The design is module oriented, thus enabling easy cascading into a multi-bit chip. The four-bit circuit dissipates 185.6 miliwatts of power and has a conversion time of 0.75 nsec. The results validate the design technology and techniques used for mixed analog-digital circuit design on a single chip
Proceedings of the Australasian Computer Science Week Multiconference
High-dimensional data is becoming more and more available due to the advent of big data and IoT. ... more High-dimensional data is becoming more and more available due to the advent of big data and IoT. Having more dimensions makes data analysis cumbersome increasing the sparsity of data points due to the problem called “curse of dimensionality“. To address this problem, global dimensionality reduction techniques are used; however, these techniques are ineffective in revealing hidden outliers from the high-dimensional space. This is due to the behaviour of outliers being hidden in the subspace where they belong; hence, a locally relevant subspace is needed to reveal the hidden outliers. In this paper, we present a technique that identifies a locally relevant subspace and associated low-dimensional subspaces by deriving a final correlation score. To verify the effectiveness of the technique in determining the generalised locally relevant subspace, we evaluate the results with a benchmark data set. Our comparative analysis shows that the technique derived the locally relevant subspace that consists of relevant dimensions presented in benchmark data set.
High-dimensional data brings new challenges and opportunities for domains such as clinical, scien... more High-dimensional data brings new challenges and opportunities for domains such as clinical, scientific and industry data. However, the curse of dimensionality that comes with the increased dimensions causes outlier identification extremely difficult because of the scattering of data points. Furthermore, clustering in high-dimensional data is challenging due to the intervention of irrelevant dimensions where a dimension may be relevant for some clusters and irrelevant for others. To address the curse of dimensionality in outlier identification, this paper presents a novel technique that generates candidate subspaces from the high-dimensional space and refines the identification of potential outliers from each subspace using a novel iterative adaptive clustering approach. Our experimental results show that the technique is effective.
This paper presents the design of ladder-type filters based on film bulk acoustic wave resonator ... more This paper presents the design of ladder-type filters based on film bulk acoustic wave resonator (FBAR) in Ku-band. The proposed FBAR filter has an insertion loss of −3 dB, out-of-band rejection of −12 dB and 3 dB bandwidth of 1.0 GHz from 15 GHz to 16 GHz. Based on the characteristics of the FBAR filter, the expected characteristics of FBAR resonators are determined by using the 1D numerical analysis. This design proves that it is possible to design a wide-bandwidth FBAR filter in Ku-band.
Wireless biosensor networks (WBSNs) collect information about biological responses and process it... more Wireless biosensor networks (WBSNs) collect information about biological responses and process it using scattered battery-power sensor nodes. Such nodes demand ultra low-power consumption for longer operating time. Ultra Wide Band (UWB) is a potential solution for WBSNs due to its advantage in low power consumption at reasonable data rate. However, such UBW technology requires high quality (Q) factor passive components. This paper presents detailed analysis, design and optimization of physical parameters of silicon-on-sapphire (SOS) and micro-electro-mechanical-systems (MEMS) inductors for application in UWB transceivers. Results showed that the 1.5 nH SOS inductor achieved Q factor of 111 and MEMS inductor achieved Q factor of 45 at 4 GHz frequency. The voltage controlled oscillator (VCO) designed with SOS inductor achieved more than 10 dBc/Hz reduction in phase noise and consumed half the power compared to VCO with MEMS inductor. Such low power VCO will improve battery life of a U...
This paper presents architecture independent characterization of embedded Java workloads based on... more This paper presents architecture independent characterization of embedded Java workloads based on the industry standard GrinderBench benchmark which includes different classes of real world embedded Java applications. This work is based on a custom built embedded Java virtual machine (JVM) simulator specifically designed for embedded JVM modeling and embodies domain specific details such as thread scheduling, algorithms used for native
In accordance with the object-oriented design principles, the flight simulation platform of each ... more In accordance with the object-oriented design principles, the flight simulation platform of each function is modular and hierarchical. The 1 Pool 2 Pieces 4 Layers software framework is proposed, using VC++ features to abstract all the replaceable modules to the same class, realizing rapid conversion of a variety of aircraft and different tests, to achieve reconfigurable. The realistic flight dynamics model considering wind disturbance is established, and with F-16 model as an example has conducted digital simulation and visual simulation. The results show that the platform has good maneuverability and fidelity, and provides reliable basis for aircraft design.
11th IEEE International Symposium on Personal Indoor and Mobile Radio Communications. PIMRC 2000. Proceedings (Cat. No.00TH8525)
This paper describes the design and performance of a mixed signal, very high speed, analog-to-dig... more This paper describes the design and performance of a mixed signal, very high speed, analog-to-digital converter implemented using gallium arsenide MESFET technology for software radio applications. The design is module oriented, thus enabling easy cascading into a multi-bit chip. The four-bit circuit dissipates 185.6 miliwatts of power and has a conversion time of 0.75 nsec. The results validate the design technology and techniques used for mixed analog-digital circuit design on a single chip
Uploads
Papers