Power Efficiency in Wireless Network Distributed Computing

Abstract

Advanced wireless applications such as sensor networks involve a close interaction between the communication and computation processes that deliver the services under stringent power constraints. Wireless network distributed computing (WNDC) is a potential solution to reducing the power consumption per node as well as that of the network. In WNDC, a computational task is executed among a network of collaborative nodes in a distributed manner as against performing the same task on a single node. In addition to providing power savings, WNDC enables power demand-supply matching that allows for system operation under a constrained power supply such as solar power. This paper presents fundamental power efficiency analysis of WNDC. The conditions for achieving power demandsupply matching and positive network power savings under power and computational latency constraints are derived. The results show the impact of non-linearity in the computational system characteristics and the communication overhead on the power savings.

MAC Protocol Implementation & Design for Wireless Distributed Computing

“…It is not feasible for a single radio device to meet these requirements but if multiple nodes form a WDC network and distribute the complex task among them these constraints can be overcome. Studies have shown that WDC has considerable benefits, particularly when the computational cost dominates over the communication overhead, which is the case when complex computational tasks are distributed among multiple network devices in short-range networks [6][7][8].…”

Task allocation and scheduling in wireless distributed computing networks

“…1, WDC workload allocation can be viewed as a problem of mapping the application's task graph to the network's communication graph such that the operational objectives are optimized [2,7]. This section presents macroscopic power and energy consumption models of individual subsystems and the WDC network, which are derived from low level power consumption models [5,8].…”

“…The time taken to process F i in N j at clock frequency f k cp is given by T ij cp ðf k cp Þ ¼ N ij cycles =f k cp , where N ij cycles denotes the number of processor cycles consumed when F i is executed in N j . Equation 2 expresses the non-linear polynomial relationship between computational power consumption P ij cp and processor's clock frequency f cp [5], where the coefficients a ij , b ij , c ij , and d ij are all positive and empirically related to the processor model and the computational task being executed.…”

“…A fundamental research question in WDC is the tradeoff between local and distributed processing [4,5] where it is important for a node to decide between the two forms of processing with regard to specific operational goals such as minimization of energy, power and time, and maximization of reliability in processing computational tasks of an application. In addition, several issues are critical to the practical implementation of collaborative applications in WDC networks, including synchronization, robust communication between nodes, WDC network control and management, willingness among participating nodes to share workload, resource allocation, fairness in load distribution in terms of energy and workload, security (trust issues and data communication over the air), and reliability in executing the application in a distributed manner (including computational accuracy and failure of slave nodes during application execution) [4,6].…”

Design and Development of Smart Brain-Machine-Brain Interface (SBMIBI) for Deep Brain Stimulation and Other Biomedical Applications

“…The use of this active technology is essentially hindering the process of miniaturization of the IPG, and making it power efficient enough to integrate it in each of the electrodes to enable discretization of the process. It is obvious that distribution of the work load implies requirement of the less resourceful hardware and software for the system, thereby enabling power efficient processing [99,100]. In addition, discretization and autonomy of the each of the electrodes will enable simultaneous operation-parallel and distributed processing of the brain activity at each of the micro brain sites-,thereby improving the temporal resolution of real-time monitoring, and offering faster response.…”