A Low-complexity, near-optimal scheduling policy for solving a restless multi-armed bandit problem occurring in a single-hop wireless network /

Download
2014
Gül, Ömer Melih
Power resources and battery lifetime are important issues for wireless networks such as wireless sensor networks (WSNs). To extend the battery lifetime, the recent advances in energy harvesting (EH) techniques propose an effective solution. EH nodes can harvest energy from environmental sources (e.g. solar, wind, vibrational, thermal) to power their sensing, computing and communication functions. In this thesis, we develop a solution to a scheduling problem under three scheduling scenarios. Firstly, we consider a single-hop wireless network where the fusion center (FC) collects data from a set of m EH nodes (e.g. nodes of a WSN). In each time slot, k of m nodes can be scheduled by the FC for transmission over k orthogonal channels. FC has no direct knowledge of battery states of nodes, or EH processes; it only has causal information of the outcomes of transmission attempts. The objective is to find a low complexity scheduling policy whereby the fusion center can collect the maximum amount of throughput in this data backlogged system, where transmission is limited by harvested energy. Energy is assumed to be stored losslessly in the batteries of nodes, up to a storage capacity (infinite capacity case is also considered). The problem is treated in finite and infinite problem horizons. Secondly, we consider the case where the infinite data backlog assumption is lifted. Thirdly, we consider a dual problem of the first scheduling problem. A low-complexity policy, UROP (Uniformizing Random Ordered Policy) is proposed, whose near optimality is shown under general energy harvesting and data arrival processes (uniform, non-uniform, independent, Markovian). Numerical examples indicate that under a reasonable-sized battery and buffer capacity, UROP uses the arriving energy and data with almost perfect efficiency. As the problem is a restless multi-armed bandit (RMAB) problem with an average reward criterion, UROP may have a wider application area than communication network.

Suggestions

Optimal transmission scheduling for energy harvesting systems and implementation of energy efficient scheduling algorithms on software defined radio /
Uçtu, Göksel; Uysal Bıyıkoğlu, Elif; Department of Electrical and Electronics Engineering (2014)
Recently, improving energy efficiency in the rapidly evolving technology field of wireless communications has become an important need. To improve the ease of use, extend the field of application of wireless communications, and sustain the applications that use the technology for longer durations with less energy expenditure, study of energy harvesting systems has gained momentum. In this thesis, an offline scheduling problem in an energy harvesting system has been solved and three scheduling alorithms have...
Path planning and localization for mobile anchor based wireless sensor networks
Erdemir, Ecenaz; Tuncer, Temel Engin; Department of Electrical and Electronics Engineering (2017)
In wireless sensor networks, sensors with limited resources are distributed in a wide area. Localizing the sensors is an important problem. Anchor nodes with known positions are used for sensor localization. A simple and efficient way of generating anchor nodes is to use mobile anchors which have built-in GPS units. In this thesis, a single mobile anchor is used to traverse the region of interest to communicate with the sensor nodes and identify their positions. Therefore planning the best trajectory for th...
A Survey on Multipath Routing Protocols for QoS Assurances in Real-Time Wireless Multimedia Sensor Networks
Hasan, Mohammed Zaki; Al-Rizzo, Hussain; Al-Turjman, Fadi (2017-01-01)
The vision of wireless multimedia sensor networks (WMSNs) is to provide real-time multimedia applications using wireless sensors deployed for long-term usage. Quality of service assurances for both best effort data and real-time multimedia applications introduced new challenges in prioritizing multipath routing protocols in WMSNs. Multipath routing approaches with multiple constraints have received considerable research interest. In this paper, a comprehensive survey of both best effort data and real-time m...
Implementation of the heed clustering protocol with sleep scheduling in tinyos 2 on a wireless sensor network testbed
Özkan, Canay; Uysal Bıyıkoğlu, Elif; Department of Electrical and Electronics Engineering (2013)
Energy efficient operation is often critical for wireless sensor networks (WSN), because network nodes operate on limited batteries. The aim of this thesis is to realize an implementation of an energy-efficient, scalable and competitive clustering protocol, HEED on MICAz motes using TinyOS 2 with further energy-efficiency enhancements. A working testbed is constructed, on which the network organizes itself to send periodic sensor measurements to a central station. In order to control overall energy consumpt...
Life time sensitive weighted clustering on wireless sensor networks
Alizadeh Jarchlo, Elnaz; Bazlamaçcı, Cüneyt Fehmi; Department of Information Systems (2013)
Wireless Sensor Networks typically include wireless sensor nodes with limited energy. Network lifetime and scalability are considered as two significant requirements for sensor network applications. In order to decrease energy consumption and increase network lifetime one can apply an efficient clustering method. The application of the clustering method proposed in this thesis (LTS-WCA) leads to reducing the energy cost and the transmission distance of each node by grouping the nodes in several clusters and...
Citation Formats
Ö. M. Gül, “A Low-complexity, near-optimal scheduling policy for solving a restless multi-armed bandit problem occurring in a single-hop wireless network /,” M.S. - Master of Science, Middle East Technical University, 2014.