Internet and Pervasive Networking

Vinton Cerf

In 2005 it is estimated that there are of the order of one billion users of the Internet and that there are at least 300 million servers. There are also many episodically connected devices, such as laptops, personal digital assistants and, increasingly, Internet-enabled mobile phones. Virtually all these devices use the Transmission Control Protocol and Internet Protocol (TCP/IP) for command/control and to carry digitized content from sender to receiver. IP does not care how it is carried or what it carries – making it the ideal neutral base on which to transport virtually anything.

There are also an increasing number of scientific instruments on the network, controllable through the network, whose output can be captured and transmitted through the Internet to storage and analysis sites. The benefit of linking these devices to the Internet lies, in part, with the ability to harness subsets of them to perform complex, collaborative tasks and to manage these instruments as a constellation to achieve specific results. The prevalence of IP has led to other layers of infrastructure, such as HTTP for the World Wide Web. Generally, people now assume that IP is virtually available and are building new layers of protocol infrastructure on top of it ("Everything on IP".

Because there are so many computers on the network, they can sometimes be harnessed, during their otherwise idle time, to carry out complex computations in parallel. A mixture of serial control and parallel computation can render a collection of computers more powerful than the most carefully planned multiprocessing system for certain, well-behaved computational tasks. So-called "GRID computing" arises from ideas like these. It should be noted that the physical locations and interconnections between the processing systems may well determine whether a computation actually experiences the kind of "speed up" one hopes for in a parallel implementation of the computation. However, physics can get in the way in the form of speed-of-light limitations (latency/propagation delay) and absolute computational capacity.

Sensor networks are also joining the Internet; and their addition, sometimes through the use of new protocols, increases the range of scientific information available. Easy archiving may allow for subsequent evaluations to be undertaken. Some of the data collected is geographically and even chronologically stamped so that the search for ways to detect trends can be enhanced. We are starting to see concurrent computations on extremely large scales, using laptops equipped to detect idle times. Epidemiology data, health care statistics, demographics are all examples of this technology to do correlated related events.

Information sharing is accelerating the pace of scientific discovery. By joining mailing lists, tracking weblogs and web pages, populating shared databases (like the human genome database), and applying advanced search tools, scientists have eminently more powerful search and correlation engines.

Radio frequency ID (identity) chips (RFID) are changing the face of inventory control, patient/person tracing, correlation of medications with patients, identification of persons or bodies subjects in death and similar uses. Generally they are passive devices that echo their identifiers when pulsed, and can be used to track shelf-life expiration and a virtually unlimited number of other applications requiring identifiers. By joining this information with global
positioning satellite data, it is possible to build a database that can be used for a variety of applications. "What things are in the neighborhood?" or "What things have been in the neighborhood" are questions that can be answered by RFID devices.

We are anticipating that billions of devices and sensors will soon be online and under control through the Internet. We can now even contemplate the extension of the Internet, through the addition of specialized protocols, to the exploration of the Solar System. By creating new space communication standards, it will be possible for future space missions (manned and robotic) to make use of the communications assets that are in place from earlier missions and still operational. The current rovers on Mars, Spirit and Opportunity are making use of this concept by relaying information back to Earth by means of the Mars Odyssey orbiter. The recently launched Mars Reconnaissance orbiter is using the new Custody File Transfer Protocol (CFDP) to test some of these concepts. We can anticipate that by the end of the decade, we may well have a two planet Internet in operation, and as further missions are launched, we can look forward to the consequent growth of an Interplanetary backbone we call the InterPlaNet.