Energy efficient network data transport through adaptive compression using the DEEP platforms

Direct measurement of event and component resolved energy dissipation in computing systems is critical for energy optimization of computing and networking applications. Prior research focuses on development of energy consumption models and custom-built energy measurement systems, but suffers from critical drawbacks. This paper addresses these limitations and presents solutions using DEEP (Decision-support for Energy Efficient Processing). DEEP is a rapidly-deployed open-source energy measurement platform architecture. The platform provides an unprecedented ability to non-intrusively measure the energy consumption associated with execution of software application code. DEEP is implemented as both an online and offline version. Evaluation demonstrates processing and energy overheads less than 1% for offline and about 5% for the online implementation. The DEEP implementation investigates the impact of data compression on network data transport. An intelligent data compression and transport algorithm is developed using the decision-support capabilities of DEEP. The algorithm creates significant energy savings (38%) in network data transport using dynamic selection of compression schemes to adapt to varying system and wireless network conditions.