Yewguan Soo

Estimation of handgrip force using frequency-band technique during fatiguing muscle contraction

In this paper, we propose a force estimation model to compute the handgrip force from SEMG signal during fatiguing muscle contraction tasks. The appropriate frequency range was analyzed using various combinations of a wavelet scale, and the highest accuracy was achieved at a range from 242 to 365 Hz. After that, eight healthy individuals performed a series of static (70%, 50%, 30%, and 20% MVC) and dynamic (0-50% MVC) muscle contraction tasks to evaluate the performance of this technique in comparison with that of former method using the Root Mean Square of the SEMG signal. Both methods had comparable results at the beginning of the experiments, before the onset of muscle fatigue. However, differences were clearly observed as the degree of muscle fatigue began to increase toward the endurance time. Under this condition, the estimated handgrip force using the proposed method improved from 17% to 134% for static contraction tasks and 40% for dynamic contraction tasks. This study overcomes the limitation of the former method during fatiguing muscle contraction tasks and, therefore, unlocks the potential of utilizing the SEMG signal as an indirect force estimation method for various applications.

Real-time video processing using native programming on Android platform

As the smartphone industry grows rapidly, smartphone applications need to be faster and real-time. For this purpose, most of the smartphone platforms run the program on the native language or compiler that can produce native code for hardware. However for the Android platform that based on the JAVA language, most of the software algorithm is running on JAVA that consumes more time to be compiled. In this paper the performance of native programming and high level programming using JAVA are compared with respect to video processing speed. Eight image processing methods are applied to each frame of the video captured from a smartphone that is running on an Android platform. The efficiencies of the two applications with difference programming language are compared by observing their frame processing rate. The experimental results show that out of the eight images processing methods, six methods that are executed using the native programming are faster than that of the JAVA programming with a total average ratio of 0.41. An application of the native programming for real-time object detection is also presented in this paper. The result shows that with native programming on Android platform, even a complicated object detection algorithm can be done in real-time.

Simultaneous measurement of force and muscle fatigue using frequency-band wavelet analysis

Static and dynamic handgrip experiments are performed in order to evaluate the effectiveness of utilizing frequency-band wavelet analysis in measuring force and muscle fatigue simultaneously. SEMG signals are recorded from flexor muscle and analyzed using continuous wavelet transform (CWT). The wavelet coefficients are grouped into high frequency (65Hz – 350Hz) and low frequency (5Hz – 45Hz) band. A significant correlation is discovered between amplitude of high frequency band and force level. On the other hand, the amplitude of low frequency band is associated with muscle fatigue. These results have an important implication for estimating force and muscle fatigue simultaneously especially during dynamic contraction.

Evaluation of frequency band technique in estimating muscle fatigue during dynamic contraction task

Long-time exposure to repetitive or monotonous work is associated with increased risk for musculoskeletal disorders that are due to muscle fatigue. Previously, researchers reported that muscle fatigue can be estimated using a low-frequency band of an SEMG signal. However, these studies were performed exclusively during static contraction tasks. The objective of the present study was to evaluate and determine the frequency range for a low-frequency band. In addition, the performance during dynamic contraction tasks was analyzed. A group of healthy university students (8 males) was recruited, and endurance handgrip tasks were conducted. SEMG signals were captured from the forearm muscle. The frequency range for the low-frequency band was redefined as 5 – 40Hz. The results from a dynamic contraction task indicated that a low-frequency band is a reliable method for indexing muscle fatigue from SEMG signals.

Quantitative estimation of muscle fatigue using surface electromyography during static muscle contraction

Muscle fatigue is commonly associated with the musculoskeletal disorder problem. Previously, various techniques were proposed to index the muscle fatigue from electromyography signal. However, quantitative measurement is still difficult to achieve. This study aimed at proposing a method to estimate the degree of muscle fatigue quantitatively. A fatigue model was first constructed using handgrip dynamometer by conducting a series of static contraction tasks. Then the degree muscle fatigue can be estimated from electromyography signal with reasonable accuracy. The error of the estimated muscle fatigue was less than 10% MVC and no significant difference was found between the estimated value and the one measured using force sensor. Although the results were promising, there were still some limitations that need to be overcome in future study.

The Relationship Between Changes in Amplitude and Instantaneous Mean Frequency at Low and High Frequency Bands During Dynamic Contraction

Surface Electromyography (SEMG) analysis of dynamic contraction is becoming more important to understand the muscle condition in real life activities. The objective of the study is to discover the relationship between the changes of amplitude and frequency at high and low frequency bands. Continuous Wavelet Transform (CWT) is utilized in order to process the SEMG data in both time and frequency domain. The result shows the potential to estimate the force level by analyzing the amplitude of high frequency band. However the influence of muscle fatigue and force variation should be identified during dynamic contraction. The proposed method provides an alternative solution for analyzing SEMG signals in different perspective.

Customized instruction set simulation for soft-core RISC processor

This paper presents the instruction set simulation process for soft-core Reduced Instruction Set Computer (RISC) processor. The aim of this paper is to provide reliable simulation platform in creating customizable instruction set for Application Specific Instruction Set Processor (ASIP). The targeted RISC processor is based on basic 8-bit PIC16C5X-compatible processor where the architecture is reconfigurable through Hardware Description Language (HDL). Instruction set architecture (ISA) has been modified in term of instruction width and machine instruction. Memory address remapping algorithm is introduced to remap the memory address to correct physical memory address due to memory banking scheme being applied. A total number of 34 instruction sets are simulated and verified its operations. Selected instruction set has been reconfigured from its original operation to demonstrate the ability to modify current instruction set to suit certain specialized application. The simulation is done using a Java-based CPU architecture simulation program and data movements at file register array and memory registers are monitored to verify the correct working operation of each instruction set. The instruction set simulation process is proved capable to be the starting point in creating a reconfigurable RISC processor with customized instruction set, inline with ASIP methodology.

Effects of rate control, bridging and antenna orientation on IEEE 802.11n link performance: An experimental analysis

In this paper we present a set of network experiments on IEEE 802.11n link constructed with open source software and commercially available hardware. Most existing works focus on optimizing MAC layer parameters in order to improve the link performance. In this paper, we investigate several “non-MAC” performance impacting factors, which are usually overlooked by the research community. The experiments were carried out in an indoor office environment over a relatively clean 5GHz band. Our experiments reveal that factors such as (i) rate control mechanism, (ii) bridging and (iii) antenna orientation can significantly impact the link performance.

Educational Development Tools for Software and Hardware Processor Design

Creativity of a computer engineer is always sought after due to the increasing demands in the production of embedded systems worldwide. To ensure teaching of computer architecture subject as interesting as possible to computer engineering students, a simulation software for understanding typical computer processors was introduced. For hardware realization, a hardware description language was also taught. In this paper, the design of a real processor was described from the specification to the actualization of the processor using the most affordable and virtually maintenance-free educational software and hardware development tools.

Feature extractor for the classification of approved Halal logo in Malaysia

This paper present a new feature extractors called the Fractionalized Principle Magnitude (FPM) that is evaluated in the classification of approved Halal logo with respect to classification accuracy and time consumptions. Feature can be classified into two group; global feature and local feature. In this study, several feature extractors have been compared with the proposed method such as histogram of gradient (HOG), Hu moment, Zernike moment and wavelet co-occurrence histogram (WCH). The experiments are conducted on 50 different approved Halal logos. The result shows that proposed FPM method achieves the highest accuracy with 90.4% whereas HOG, Zernike moment, WCH and Hu moment achieve 75.2%, 64.4%, 47.2% 44.4% of accuracies, respectively.