Mobile nodes have limited computing resources, among which battery is one of the most important resources, since the lifetime of a node depends on the amount of battery and energy usage in the node. Application processes on mobile nodes include those are important for the users for certain duration of time, and those not. Therefore, exhausting whole battery for running non-important ones such as games, would be waste of energy. To maximize the user satisfactions with limited amount of energy, a sophisticated pacing mechanism is required for energy usage. We propose a process lifetime reservation system, called “pSurvive.” It enables users to reserve energy for running specific processes for a certain duration of time. Users are allowed to run any applications until the total energy consumption reaches the “deadline” to run the reserved processes. pSurvive enables this with the following three novel mechanisms. First, P-Monitor monitors running processes and devices (e.g. NIC, LCD, etc.) in a node for their energy consumption. Second, P-Analyzer estimates future energy consumption based on the energy usage information given by P-Monitor. Third, P-Reserver accepts energy reservation requests from users, and pace the energy consumption by shedding unnecessary tasks. This paper first discusses requirements for energy reservation on mobile nodes. It then reports the design and the implementation of pSurvive for Android mobile nodes, and shows that pSurvive achieves process lifetime reservation based on the fine-grained energy monitoring.

Life rhythm analysis for elderly people has recently been paid mach attention in many countries with long life expectancy. Wireless sensors attached to a human body are considered useful in automating the life rhythm analysis. We propose a system of such a monitoring and analysis system. This system has two advantages: unconscious monitoring and continuous analysis. We receive data from a mule with wireless sensors. The mule consists of pressure sensors connected with Mica2 Motes and sensed data is analyzed to extract a user's life rhythm.

Extracting human's health condition using wireless sensors impose a challenge in the balance between data storage and CPU power. Taking available resources on nodes into consideration, we apply Step-wise Context Extraction (SCE) to AoK mule system [4]. SCE compresses data obtained at the mule into a context as the data is forwarded. In this demonstration, we show how the raw data is converted into the context in the AoK system.