Low Power Considerations in Ubiquitous Computing

Introduction

Ubiquitous computing is emerging as one of the biggest applications for computing systems and carries with it a huge potential for use in every walk of life. There are many ubiquitous systems in use right now that have the potential to develop into complex systems that are more context aware. For example, most commercial cars have cruise control. Cruise control detects the current condition of the car, its speed, and quietly makes adjustments to the throttle to match a preset speed (Figure 1). This is similar to how ubiquitous systems should work. Though a simple concept on the surface, the idea of cruise control opens up a whole new field of possibilities regarding context awareness. With the widespread use of global positioning systems, a simple cruise control system can be enhanced with context awareness by contacting a global database to retrieve local traffic information. With the use of a GPS system, the preset speed then becomes dynamic, and can adjust itself automatically to fit the local speed limit. This system can also be used to detect reduced speed zones like school zones and construction areas. With an even more involved database, there exist a possibility of the cruise control system being able to synchronize with traffic lights to advise the driver when to come to a stop in order to avoid potential traffic accidents or tickets.

Figure 1.

Example of ubiquitous computing in a cruise control system of a car

This example describes a high-end application of ubiquitous computing that is very power-hungry in design. However, this application can rely on the car’s battery to act as a big power source, satisfying its power needs. There are a myriad of other possible applications of ubiquitous computing, such as sensors and embedded computers, which do not share such resources. These devices do not have access to large power supplies and instead must operate within the power budget set by small portable batteries.

Thus, power consumption imposes one of the biggest restrictions on the development of ubiquitous computing. The reason for this is that ubiquitous computing systems must be designed to be compact and self-sufficient, which will restrict the amount of resources that can be utilized by the system. Ideally, maintenance levels have to be kept at a minimum and energy costs reduced greatly. If the system is designed to be mobile, then it is expected that the only available power source will be a battery. Subsequently, size restrictions will limit the performance of the battery and the amount of power consumption in the system. Unfortunately, there are also large sources of energy waste present in modern day computing systems, which hinder the application of ubiquitous computing for every walk of life.

The sources that cause this energy waste are at both the hardware and software level. On the hardware level, circuits generate heat and dissipate power. On the software level, operating systems and applications increase in complexity as well as computational cost. For every level of a computing system, there exist some set of problems that cause the unnecessary expenditure of energy. Even newly developed technologies, such as wireless communication, open up new avenues for expensive operations that need to be performed on embedded and distributed systems.

It is generally accepted that as computing systems become more powerful, energy demands will rise. Whereas the traditional mainframe and desktop approaches focus heavily on performance, the ubiquitous computing paradigm must look to balance performance with cost. There are many techniques that can be employed to achieve energy efficiency. For example, circuits can be redesigned to operate at a lower voltage supply. Software systems can be designed to run more efficiently by scheduling for active and inactive periods of time. These systems can even be scaled back in most of its design aspects to accomplish the same task with less overhead costs.

This chapter will approach these problems by first taking an overview of the many sources of energy waste in ubiquitous computing systems, and will cover many power issues in ubiquitous computing ranging from transistor design to the user application layer. Then, there will be a look at the field of low power design that presents a large array of solutions that are being explored to reduce power consumption on a range of computing devices. Then, operating system aspects of low power designs will be explored in greater depth. Lastly, there will be a look at the latest research developments.