An Energy-driven Design Methodology for Distributing DSP Applications (10 pages)

Previewing pages 1, 2, 3 of 10 page document View the full content.
View Full Document

An Energy-driven Design Methodology for Distributing DSP Applications



Previewing pages 1, 2, 3 of actual document.

View the full content.
View Full Document
View Full Document

21 views

Unformatted text preview:

In Proceedings of the IEEE Real Time Systems Symposium Tucson Arizona December 2007 An Energy driven Design Methodology for Distributing DSP Applications across Wireless Sensor Networks Chung Ching Shen William Plishker Shuvra S Bhattacharyya and Neil Goldsman Dept of Electrical and Computer Engineering and Institute for Advanced Computer Studies University of Maryland at College Park USA ccshen plishker ssb neil umd edu Abstract Wireless sensor network WSN applications have been studied extensively in recent years Such applications involve resource limited embedded sensor nodes that have small size and low power requirements Based on the need for extended network lifetimes in WSNs in terms of energy use the energy efficiency of computation and communication operations in the embedded sensor nodes becomes critical Digital signal processing DSP applications typically require intensive data processing operations They are difficult to apply directly in resource limited WSNs because their operational complexity can strongly influence the network lifetime In this paper we present a design methodology for modeling and implementing DSP applications applied to wireless sensor networks This methodology explores efficient modeling techniques for DSP applications including acoustic sensing and data processing derives formulations of energy driven partitioning for distributing such applications across wireless sensor networks and develops efficient heuristic algorithms for finding partitioning results that maximize the network lifetime A case study involving a speech recognition system demonstrates the capabilities of our proposed methodology 1 Introduction and related work In a hierarchical wireless sensor network 11 sensor nodes are clustered into groups and their roles are divided into master e g the cluster head and slave nodes for more efficient structuring of network traffic The master node is typically fully featured that is the platform is equipped with a relatively high performance processor and transceiver larger size memory storage and sizable energy source Conversely slave nodes are lean in terms of features the platforms are equipped with simple processors e g small microcon trollers simple transceivers sensors limited memory storage and relatively small energy resources Thus slave nodes are typically equipped to carry out simple computations and transmit only the required processed data to the associated master nodes for more computationally intensive tasks In this paper the targeted wireless sensor network structures are such single hop master slave topologies Such static configurations of network structure have simple routing characteristics and protocol demands and allow designers to optimize effectively for network lifetime in terms of energy consumption as well as for scalability through hierarchical organization Digital signal processing DSP applications are often relevant to processing sensor data and usually require intensive computation The behavior of many DSP applications can be characterized with regular computation patterns and modeled efficiently through dataflow graphs By analyzing a well designed dataflow model of an application operational efficiency can be estimated and optimized accordingly e g see 2 12 In a typical WSN configuration resource limited slave nodes are unable to handle computationally intensive tasks due to lack of hardware and software support In such a case microcontrollers in slave nodes can perform relatively light weight computations With advances in integrated circuit technology slave nodes can be equipped with increasing amounts of computational resources such as digital signal processor subsystems Then the microcontrollers can perform protocol and control tasks while the DSP processors perform more intensive computational tasks The energy consumption of the nodes in a wireless sensor network must be carefully optimized to increase network lifetime Computation and communication tasks that sensor nodes execute affect major energy consumed Especially communication tasks on the transceiver dominate overall power consumption on a sensor node 5 Thus the amount of data to be trans mitted across the wireless channel should be minimized due to the reduction of the turn on time of the transceivers For distributing DSP computations to the WSN this reduction is significant since DSP applications usually process a large amount of data for their tasks In this paper we present an algorithm that addresses this objective for DSP applications that are modeled as dataflow graphs Specifically our algorithm finds an efficient trade off between the workloads of computation and communication tasks in both master and slave nodes Many useful approaches have been studied previously to reduce the energy consumption of sensor nodes In 15 Singh discusses system level trade offs related to energy costs of WSN technologies Shih 14 distributes the FFT function over a master node and associated slave nodes to reduce energy consumption Kumar 11 explores energy and latency trade offs by considering different computational capabilities for master and slave nodes Wang 16 develops an approach that partitions applications between master and slave nodes and also applies dynamic voltage scaling to further reduce power consumption In contrast to Wang s approach the partitioning method presented in this paper applies coarsegrain analysis of dataflow graphs as well as integration within a dataflow based DSP design tool This tool called the DIF dataflow interchange format package is introduced in 8 Ko 10 introduces the general approach of reducing data traffic across WSN nodes by determining and exploiting the lowest data token delivery points within an application dataflow graph This paper builds on the approach and problem formulations introduced by Ko and develops an efficient heuristic method that is more efficient and scalable compared to the exhaustive search approach employed by Ko We also go beyond the developments of 10 in this paper by addressing both homogeneous and heterogeneous network organizations taking into account energy consumption in more detail as well as real time latency constraints based on measured and simulated task profiles and as mentioned above by integrating our partitioning methods into the DIF package 2 Design flow overview 8 DIF is a unified textual language for expressing different kinds of dataflow semantics The DIF package is a software package


Access the best Study Guides, Lecture Notes and Practice Exams

Loading Unlocking...
Login

Join to view An Energy-driven Design Methodology for Distributing DSP Applications and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view An Energy-driven Design Methodology for Distributing DSP Applications and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?