An MDP-based Application Oriented Optimal Policy for Wireless Sensor Networks Arslan Munir and Ann Gordon-Ross ∗ Department of Electrical and Computer Engineering University of Florida, Gainesville, Florida, USA amunir@ufl.edu, ann@chrec.org ABSTRACT Technological advancements due to Moore’s law have led to the proliferation of complex wireless sensor network (WSN) domains. One commonality across all WSN domains is the need to meet application requirements (i.e. lifetime, respon- siveness, etc.) through domain specific sensor node design. Techniques such as sensor node parameter tuning enable WSN designers to specialize tunable parameters (i.e. proces- sor voltage and frequency, sensing frequency, etc.) to meet these application requirements. However, given WSN do- main diversity, varying environmental situations (stimuli), and sensor node complexity, sensor node parameter tuning is a very challenging task. In this paper, we propose an auto- mated Markov Decision Process (MDP)-based methodology to prescribe optimal sensor node operation (selection of val- ues for tunable parameters such as processor voltage, pro- cessor frequency, and sensing frequency) to meet application requirements and adapt to changing environmental stimuli. Numerical results confirm the optimality of our proposed methodology and reveal that our methodology more closely meets application requirements compared to other feasible policies. Categories and Subject Descriptors C.4 [Computer Systems Organization]: Performance of Systems—Design studies, modeling techniques ; C.3 [Computer Systems Organization]: Special-Purpose and Application- Based Systems—Real-time and embedded systems General Terms Design, Performance Keywords Wireless sensor networks, dynamic optimization, MDP ∗ Also with the NSF Center for High-Performance Reconfig- urable Computing (CHREC) at the University of Florida, Gainesville, Florida, USA. Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. CODES+ISSS’09, October 11–16, 2009, Grenoble, France. Copyright 2009 ACM 978-1-60558-628-1/09/10 ...$10.00. 1. INTRODUCTION AND MOTIVATION Advances in silicon technology due to Moore’s law have led to the proliferation of increasingly capable wireless sen- sor networks (WSNs) in application domains such as secu- rity and defense systems, industrial monitoring, building au- tomation, logistics, ecology, environment and ambient condi- tions monitoring, health care, home and office applications, vehicle tracking, etc. However, this wide application diver- sity combined with increasing complexity, functionality re- quirements, and highly constrained operating environments makes WSN design very challenging - even described as re- quiring “2.5 Ph.D’s” [8]. One critical WSN design challenge involves meeting ap- plication requirements such as reliability, lifetime, through- put, delay (responsiveness), etc. for myriad of application domains. For example, a vineyard irrigation system may re- quire less responsiveness to environmental stimuli (i.e. de- creased irrigation during wet periods), but have a long life- time requirement. On the other hand, in a disaster relief ap- plication, sensor nodes may require high responsiveness but have a short lifetime. Additional requirements may include high adaptability to rapid communication network changes as sensor nodes are destroyed. Meeting these application specific requirements is critical to accomplishing the appli- cation’s assigned function and satisfying these demands in a scalable and cost-effective way is a challenging task. Commercial off-the-shelf (COTS) sensor nodes have diffi- culty meeting application requirements due to inherent man- ufacturing traits. In order to reduce manufacturing costs, generic COTS sensor nodes capable of implementing nearly any application are produced in large volumes, and are not specialized to meet any specific application requirements. In order to meet application requirements, sensor nodes must possess tunable parameters. Fortunately, some COTS have tunable parameters such as processor voltage, processor fre- quency, sensing frequency, radio transmission power, and radio transmission frequency, etc. Sensor node parameter tuning is the process of determin- ing appropriate parameter values which meet application requirements. However, determining such values presents several tuning challenges. First, application managers (the individuals responsible for WSN deployment) typically lack sufficient technical expertise [8], [6], as many managers are non-experts (i.e. biologists, teachers, structural engineers, agriculturists, etc.). In addition, parameter value tuning is still a cumbersome and time consuming task even for expert application managers due to unpredictable WSN environ- ments and difficulty in creating accurate simulation envi- 183