Enabling Energy-Efficient Wireless Sensing with Improved Service Quality

04. November 2016

Organisator:  Mohamed Abdelaal, M.Sc.

Ort:  A5 1-159

M i t t e i l u n g 

Im Rahmen der Disputation seines Promotionsverfahrens hält Herr Mohamed Abdelaal, M.Sc., seinen öffentlichen Vortrag zum Thema

„Enabling Energy-Efficient Wireless Sensing with Improved Service Quality“

am Freitag, den 4. November 2016, um 14:00 Uhr im A5 1-159.

Abstract: 

Wireless Sensor Networks (WSNs) are spatially distributed autonomous sensor nodes (SNs) over the sensing field to monitor physical or environmental conditions, such as temperature, sound, pressure, etc. Thereupon, measurements are conveyed in a cooperative manner to the application manager via a sink node. In general, WSNs have some features that differentiate them from other wireless networks including 1) deployment in harsh environments and 2) strong restrictions on hardware and software capabilities in terms of processing speed, memory storage, and energy supply. Such sensors usually carry limited, irreplaceable energy resources. Therefore, lifetime adequacy is a significant feature of all WSNs.

The crux behind this work is to break the "downward spiral" between reducing certain quality-of-service (QoS) measures to a degree still tolerable by the application (such as, for example, precision and latency) and extending the application's lifetime. In fact, the WSN literature is rich of endeavors in the field of energy efficiency and QoS control. For the sake of identifying research challenges and possible gaps, we conduct a comprehensive survey of the state-of-the-art. The reported methods have been classified into node-oriented methods, data-oriented methods, and network-oriented methods. The impact of each method on the provided service quality is briefly discussed. Furthermore, we report on QoS control methods in WSNs. Based on this survey, it was obvious that several single- and multi-objective optimization methods have been proposed to handle this relationship. In fact, single-objective methods are not a practical solution in most WSN applications in which many QoS parameters are deliberately engaged. The multi-objective optimization (MOO) methods provide reasonable solutions but they suffer -- in most cases -- from poor scalability and lack of flexibility against run-time dynamics.

Alternatively, the thesis introduces an energy-centric strategy to manipulate the energy-QoS relationship in the light of the WSN application relevant context information. As its name implies, it is a design procedure in which the amount of saved energy -- during run-time -- is dynamically adjusted according to changes in the application and environmental conditions. These adjustments provide the required energy to enhance the provided service qualities. We propose exploiting design-time knowledge to leverage self-adaptive mechanisms. For example, knowing the application scenario helps in estimating the minimum amount of energy, required to achieve the expected lifetime. Hence, the sensor nodes can freely adjust their energy-saving method provided that they possess the predefined minimum energy level. Due to diversity of WSN applications, we introduce three different methods, based on the energy-centric strategy, namely Fuzzy transform-based precise data compression, reliable virtual sensing, and lifetime planning.

In the first method, the radio energy consumption, as a dominant energy-consumer, is reduced via multi-objective data encoding. The literature has some lossy compressors such as discrete wavelet transform (DWT), discrete cosine transform (DCT), and lightweight temporal compression (LTC) methods. However, those methods suffer from drawbacks, such as being ill-suited for resources-taxed devices and having weak immunity against possible outliers. These limitations motivate toward exploring a recently-developed Fuzzy transform to be used for sensor data compression. Actually, the Fuzzy compressor shows a comparable precision performance with the aforementioned methods. However, further improvement of the recovery precision via adapting the transform is sought. Then, a modified version, referred to as FuzzyCAT, has been proposed. FuzzyCAT has high compression ratios and precision via detecting the input signal curvature and dynamically modifying the transform's approximating function.

In the second method, virtual sensors (VSs) are proposed as a novel technique for reducing the excessive energy consumed by some sensors (such as GPS and Gas sensors); and simultaneously slashing the event-miss probability. Generally, VSs are orchestrations of HW/SW components whose output describes a certain phenomenon. Such a phenomenon can be directly acquired through adopting an "energy-hungry" sensor. The method has been evaluated through two case studies including object tracking and gas leaks detection. In both cases, the lifetime of the main sensors has been significantly extended. Moreover, reliability of the VSs is improved via adopting ontology-based generated rules for sensor selection (sensing quality and environmental conditions are the selection criteria).

In the third method, a novel concept is developed for the interplay between energy efficiency and QoS requirements. Instead of maximizing the lifetime, we can only meet the application-expected lifetime, and simultaneously improve the provided QoS. A self-adaptive framework has been proposed to respond to the environmental dynamics. Moreover, a hierarchical monitor-analyze-plan-execute (MAPE) architecture has been introduced to form a global control loop. The results show that lifetime planning highly improves the QoS characteristics. This profit came at the expense of reducing the WSN lifetime. But the resultant lifetime, shortened lifetime is still long enough to complete the assigned WSN task.

Mit der Teilnahme von Zuhörerinnen und Zuhörern an der anschließenden Prüfung bis 15:30 Uhr ist Herr Abdelaal einverstanden.

gez. Prof. Dr.-Ing. A. Hahn