JMM Abstracts 

Vol.12 No.3&4 April 30, 2017

A Scheduling Method for Switching Playback Speed in Selective Contents Broadcasting (181-196)
       
Daichi Fukui and Yusuke Gotoh
In selective contents broadcasting, i.e. watching contents users selected themselves, the server can deliver several contents to many users. However, when users watch the data continuously, waiting time occurs by decreasing the available bandwidth and increasing the number of contents. Therefore, many researchers have proposed scheduling methods to reduce the waiting time. Although the conventional method reduces waiting time by producing the broadcast schedule in fast-forwarding, that for playing contents in normal playback becomes lengthened. In this paper, we propose a scheduling method for switching the playback speed in selective contents broadcasting. Our proposed method can make the broadcast schedule based on the configuration of the program and the available bandwidth. In addition, waiting time can be reduced by dividing each content into two types of data for fast-forwarding and normal playback and scheduling them.

Intelligent Personal Health Devices Converged with Internet of Things Networks (197-212)
       
James J. Kang and Henry Larkin
Smartphone technology has become more popular and innovative over the last few years, and has led to the prevalence of wearable devices embedded with body sensors for fitness tracking and various smartphone features. Internet of Things (IoT), which can interact with wearables and personal sensor devices (PSDs), is emerging with technologies such as mobile health (mHealth), the cloud, big data and smart environments like smart homes. It may also provide enhanced services utilising health data obtained from physiological sensors. When these sensors are converged with IoT devices, the volume of transactions and traffic are expected to increase immensely due to the increased demand of health data from the IoT network. These additional demands will affect the existing mHealth services. Health service providers may also demand more data to enhance their services such as real-time monitoring and actuation of sensors alongside the existing monitoring of traffic. Both of these situations can cause rapid battery consumption and consume significant bandwidth. Some PSDs are implanted on or inside the body, and may require invasive surgical operations to replace batteries, such as for a heart pacemaker. It is therefore crucial to save and conserve power consumption in order to reduce the frequency of such procedures as well as health data transmission when needed. There has not yet been any research into managing and controlling data processing and transmission to reduce transactions by applying intelligence onto body sensors. This paper provides a novel approach and solution to reduce data transactions in sensors and allow for the transfer of critical data without failure to medical practitioners over IoT traffic. This can be done via an inference system to transfer health data collected by body sensors efficiently and effectively to mHealth and IoT networks. The results from the experiments to reduce bandwidth and battery resources with heart rate sensors show a possible savings in resource usage of between 66% and 99.5%. Battery power can be saved by 3.14 Watts in the experiments if the transmission of a single 1KB data point is reduced, and by 7.47 Watts if the transmission of 628 data points totalling the size of 120KB is reduced. The accuracy of data inference between the originally sensed data and the data transmitted after inference can be maintained by up to 99% or more. Such savings have the potential of making always-on mHealth devices a practical reality. This research contributes a low-overhead approach to mHealth sensors by inferring the processing and transferring of data.

When Game Theories Meets Security and Privacy Related Risk Assessment of Vehicular Networks (VANET) (213-224)
       
Sara Bahamou, Driss El Ouadghiri, and Jean-Marie Bonnin
Vehicular ad-hoc networks faces numerous challenges, in particular Security and Privacy issues. Here, security assessment, risk evaluation and many other security features/ calculations are based generally on theoretical ideas, that may project in different level real life conditions. Hence, the existing of an important risk that these analysis may not work properly. We shouldn't forget that the important goal of VANET applications deployment is to reduce accident rate, which means the reduction of the human deaths and injuries number. Hence, our focus on the importance of strengthening the security aspect of different VANET applications and services, Which could be achieved by relying on security requirements. We found many research approaches covering a significant part of VANETs security and risk assessment challenges. This article is dedicated to the most crucial challenges in every security analysis methodology which is the study of different attack scenarios. These scenarios are generally in accordance with the most feared threats. So our goal here, is to determine the most probably attack scenario and how we can deploy the appropriate countermeasures in such a way some risks are acceptable. Therefore, our approach to VANET defense applies attack tree- defense tree for advanced vulnerabilities assessment and intrusion detection. Attack tree processing shows how attackers can penetrate a VANET, so by identifying critical vulnerabilities we can provide strategies for protecting the critical network assets. Thus, attack-tree leads us to the optimal intrusion detection and attack response. A strong link is existed between attacker and defender assets, which can be modeled by a mathematical analysis, called game theory analysis. This game approach has demonstrated the interaction in the attack-tree processing.

Comparing the Placement of Two Arm-Worn Devices for Recognizing Dynamic Hand Gestures (225-242)
       
Kathrin Kefer, Clemens Holzmann, and Rainhard Dieter Findling
Dynamic hand gestures become increasingly popular as touch-free input modality for interactive systems. There exists a variety of arm-worn devices for recognition of hand gestures, which differ not only in their capabilities, but also in their positioning on users' arms. These differences in positioning might influence how well gestures are recognized, leading to different gesture recognition accuracies. In this paper we therefore investigate the effect of device placement on dynamic hand gesture recognition accuracy. We consider devices being strapped to the forearm on two positions: the wrist and below the elbow. These positions represent smart watches being worn on the wrist and devices with EMG sensors for the detection of static hand gestures (e.g\ spreading the fingers) being worn right below the elbow. Our initial hypothesis is that wrist-worn devices will have better recognition accuracy, caused by higher acceleration values of a bigger action radius with dynamic hand gestures on the wrist. We conduct a comparative study using an LG G Watch and Thalmic Labs' Myo armband, for which we record a total of 12960 gesture samples of eight simple dynamic gestures in three different variants with eight participants. We  evaluate a potential difference in gesture recognition accuracies using different feature sets and classifiers. Although recognition accuracies for wrist-worn devices seem higher, the difference is not statistically significant due to substantial variations in accuracy across participants. We thus cannot conclude that different positions of gesture recording devices on the lower arm have significant influence or correctly recognizing arm gestures.

A Comparison of Techniques for Cross-Device Interaction from Mobile Devices to Large Displays (243-264)
       
Jeni Paay, Dimitrios Raptis, Jesper Kjeldskov, Bjarke M. Lauridsen, Ivan S. Penchev, Elias Ringhauge, and Eric V. Ruder
In recent years there has been an increasing interest in cross-device interaction research involving mobile computing. We contribute to this research with a comparative study of four interaction techniques for moving information from a mobile device to a large display. The four techniques (Pinch, Swipe, Throw, and Tilt) were compared through a laboratory experiment with 53 participants, measuring their effectiveness, efficiency and error size. Findings from the experiment revealed that the Swipe technique performed best on all measures. In terms of effectiveness, the Tilt technique performed the worst, and especially so with small targets. In terms of efficiency and error size, the Pinch technique was the slowest and also the most imprecise. We also found that target size mattered considerably for all techniques, confirming previous research. Based on our findings we discuss why the individual techniques performed as observed, and discuss implications for using mobile devices in cross-device interaction design.

Sparse Canonicial Correlation Analysis for Mobile Media Recognition on the Cloud (265-276)
        Y
angjiang Wang, Bin Zhou, Weifeng Liu, and Huimin Zhang
With the rapid development of the Internet technology and smartphone, people can easily capture and upload media information including text, audio, photos, and video. And then it becomes one critical demand to effectively and efficiently manage these personal multimedia that are often presented in multiple modalities. Canonical correlation analysis (CCA) has been widely employed for multi-modal data in many applications because of its promising performance in feature extraction and subspace learning for multivariate vectors. However, the traditional CCA may be difficult to interpret especially when the original variables are expected to involve only a few components. In this paper, we develop a mobile media recognition method on the cloud. Particularly, we propose sparse canonical correlation analysis (SCCA) on the cloud. SCCA can find a reasonable trade-off between statistical fidelity and interpretability. Furthermore, we employ a generalised power method to optimise the SCCA algorithm. Finally, we conduct extensive experiments for recognition on several popular databases including UCI datasets and USAA dataset. Experimental results demonstrate that the proposed SCCA algorithm outperforms the traditional CCA algorithm.

A Variance Distortion Rate Control Scheme for combined Spatial-Temporal Scalable Video Coding (277-290)
       
L. Balaji, A. Dhanalakshmi, and C. Chellaswamy
Rate Control plays an important role in video compression for transportation over heterogeneous network bandwidth varying conditions. A combined spatial-temporal rate control scheme is proposed for scalable video coding. Introductory quantization parameter estimation is determined based on complete variance distortion method for I frame of first GOP and estimates buffer occupancy level. With the estimated buffer level, target bits are determined considering the coding complexity is proposed in the rate control scheme. In addition, a proportional integral and derivative (PID) controller that calculates the error and minimize fluctuation between the actual buffer fullness and target buffer fullness for competent buffer utilization. As a result, the proposed scheme exploits entire buffer exclusive of crossing overflow and underflow level. The investigational results are compared with other two benchmark schemes and the proposed scheme can able to achieve better target bit adjustment with condensed fluctuations and competent buffer utilization.

Analysis and Evaluation of Feature Detection and Tracking Techniques Using Open CV with Focus on Markerless Augmented Reality Applications (291-302)
       
Gustavo Maglhaes Moura and Rodrigo Luis de Souza da Silva
Augmented Reality (AR) is a technology able to extend human interactions with the real world. One field of study in AR is the use of real objects as markers. To perform this task, feature recognition of the real world by computer systems must be performed. This work consists in the analysis and evaluation of several algorithms available in OpenCV library that allow the detection of pre-established patterns in images and videos. The main contribution of this work is to present the most appropriate combination of algorithms to help the development of markerless AR applications.

Back to JMM Online Front Page