Software - Pervasive Computing

Signs that show the way to the nearest cinema; cars that automatically locate the next parking space; shirts that tell the washing machine what temperature they should be washed—in the future, according to Prof. Alois Ferscha, objects and people will be enveloped in their very own "digital auras." This might sound esoteric, but Ferscha, who is the Director of the Institute of Pervasive Computing at the Johannes Kepler University in Linz, Austria has some very concrete ideas of what it might entail. "We create an artificial aura for people and objects by fitting them, as it were, with a digital cloak. Whenever two such auras come into contact, information flows." In other words, coded preference profiles are exchanged and compared. For example, if your profile authorizes this, an electronic movie poster might transmit the trailer of the latest box office hit to your PDA; or your cell phone might inform you that the woman sitting at the next table in the cafe wants to sell her car.

The technology is already available. The Linz researchers have fitted various objects with so-called RFID tags—small chips that store relevant data. Communication takes place via the Bluetooth wireless radio standard. Initial demonstrations already exist in the fields of healthcare, the home, and traffic management. "The major challenges now are to write universally applicable digital auras for a huge number of people and things, to ensure that these auras can change over time, to transfer and compare them wirelessly, and, finally, to use washable microchips that can be integrated into clothing," explains Ferscha. For the last three years, his institute has been working closely on the digital aura project with Dr. Lothar Borrmann and others at Siemens Corporate Technology’s Software Architecture department, a part of the Siemens Software & Engineering division.

No Keyboard, No Mouse. Pervasive Computing (PvC)—also known as Ubiquitous Computing (see Pictures of the Future, Fall 2002,  "Ubiquitous Computing")—will usher in a new era. Instead of do-it-all computers, we will see the advent of simple, task-specific, miniaturized and intuitively operable processors that will be invisibly integrated in everyday objects. Similarly, traditional input devices such as keyboards and mice will not be required. Instead, the processors will be controlled by electronic, optical, acoustic or chemical sensors, and they will output via actuators such motors or other control units.

In order to reach that point, however, researchers need to develop new software that is capable of the following:

•  Self-configuration, that is, automatic adaptation to changing environments

•  Self-optimization, including continual monitoring and analysis of its own performance and the use of available resources according to specific processes

•  Self-organization and the implementation of decisions across the system as a whole

•  Self-protection, meaning identification and control of unauthorized access and virus activity

•  Self-repair, for example, discovering and resolving problems

•  Self-teaching, that is, recognition of behavioral patterns and their incorporation in internal management mechanisms. Of particular importance here is sensitivity to context.

In other words, the system must not only be capable of recognizing objects and persons, but it must also be able to prepare for future situations. Although science and industry are still at the start of this project, two things are already evident: Classic programs that process only one predetermined task are outdated; by contrast, the PvC environment requires global platforms and software that integrate individual systems such as mobile devices, sensor networks and applications in vehicles or intelligent homes.

"When building a house, it’s not so important whether you use brick, stone or wood. Similarly, the important thing here is not a specific programming language but rather the right software architecture," explains Ferscha. For this reason, his team has developed a new architecture that consists of three levels. The top level is where sensor data such as temperature, humidity, and pulse frequency are received from the immediate environment. The middle level is where data for the specific application are processed and converted into a form that can be understood by the various embedded systems. The preferred language is XML (Extensible Markup Language)—a universal and extensible data-description language that is independent of any specific platform or operating system. In fact, this is why the Linz team has used it to code the personal profile contained in a digital aura. Such a profile features not only a personal description including name and address but also changeable data such as favorite music and, above all, personal preferences and intended courses of action. In turn, it is crucial that these are evaluated according to context, since a person’s favorite music may differ between morning and evening. Similarly, readiness to engage in small talk is probably higher when relaxing in a bar than at work in the office.

MOPS Lay the Ground Rules. Finally, certain rules control how the mini-processors and networks should behave. "This is the lowest level of the architecture," says Ferscha. By helping to control motors, displays and even complete networks, such rules—also known as "policies"—enable the system to operate autonomously. "These rules are required to control complexity. They center on a component’s subject, target object, event, condition and action. You can store them in a database or decentrally," says Christoph Niedermeier of CT’s Software & Engineering Architecture department. MOPS (Mobility cOmmunication & Policy-based Systems) is the name of the project that Niedermeier is running in cooperation with scientists from the Ludwig Maximilian University in Munich. The project’s objective is to develop management policies for fourth generation mobile radio networks. An additional MOPS feature will be that software problems will be treated remotely and updates downloaded from the network. By mid-2005 Niedermeier’s team expects to produce a demonstration that simulates the policy-based control of software downloads to a large number of terminals.

In the light of such questions, it is evident that PvC will be implemented only in certain spheres (see boxes) for the foreseeable future. Nevertheless, common standards must be adopted. To date, however, industry has failed to agree on appropriate wireless protocols or language and software architectures for this area. "The subject is simply too complex, and companies are still orientating themselves. Many approaches are being tried out at present," says Dr. Michael Berger from the Intelligent Autonomous Systems department at CT’s Information & Communications Group. The department develops PvC solutions for Siemens Com and Siemens VDO.

Why Standards Are Needed. The need for standards is especially evident in the networking area. Although an infrastructure has existed for quite some time here, uncertainty still remains as to which wireless standards will prevail (see  Software Trends). According to the authors of a study at TA-Swiss, the Center for Technology Assessment in Bern, Switzerland, "Many devices will probably support the IPv6 Internet protocol. For network applications, however, additional protocols for dispersed architectures will be required. While there are numerous systems for this purpose, these are still proprietary and not interoperable. The development of manufacturer-independent standards is overdue and hinders the breakthrough of appropriate systems." Nevertheless, Alois Ferscha is confident that PvC will become a reality. "The development of network capability was the first step. The next step will be to create ubiquitous networked computer systems that can recognize situations and people’s intentions, and support them in the best possible way," he says.

Evdoxia Tsakiridou