2.3 Framework for Technology Creation

2.3 Framework for Technology Creation

The incremental evolution of a wireless Internet, described in Section 2.2, takes a bottom-up approach of augmenting existing wireless technology and Internet protocols in order to provide a smoother interface between the two marriage partners. On the wireless side, the principal aim is to increase channel bit rate. On the Internet side, extended protocols aim to accommodate mobility, the variable quality of wireless signals, and vulnerability of wireless systems to eavesdropping and unauthorized access. In this section, we introduce a top-down approach that aims for a wireless Internet that is more than the sum of the two existing communications systems. This approach begins with a three-dimensional analysis, including the characteristics of (1) the endpoints of communication, (2) the information transferred, and (3) the physical nature of wireless signals. In this analysis, we find significant differences from the wired Internet on all three dimensions. The common aspect of all three dimensions can be summed up in the word "geography." Cellular systems aim to be the same "anytime, anywhere," and the name "World Wide Web" carries a similar suggestion. By contrast, our analysis of wireless Internet requirements in the following paragraphs reveals a fundamental dependence on location, including (1) locations of information terminals (geography of users), (2) the location-dependent relevance of information (geography of information), and (3) location-dependent quality of signals (geography of signal transmission). Section 2.4 refers to examples of research in progress that aligns technology with the geography of users, the geography of information, and the geography of signal transmission.

2.3.1 The Geography of Wireless Internet Users

The Internet was originally designed to move data packets carrying many types of information between host computers in stationary, known locations. By contrast, cellular networks were originally designed to carry telephone calls and short messages in systems that are matched to the geographical distribution of subscribers, their mobility patterns, and the temporal distribution their service needs. Technology creation and deployment are considerably more complicated in a wireless Internet because mobile terminals with different capabilities will transmit and receive multimedia information in a variety of formats, with widely different quality-of-service requirements that place varying demands on network resources.

Wireless Internet technology needs to be sensitive to the characteristics of the sources and destinations of information, which will often be groups that share information. Groups form and dissolve as clusters in time and space. The formation and disintegration of such groups may or may not be initiated by the users involved. Key characteristics of the geography of users are location, mobility state (speed and direction), timing of information needs, and demographics of individuals and user groups. An example of a group formed spontaneously is the population of mobile callers in an unexpected traffic jam. In this case, the defining characteristics of the group are the locations and mobility states of the group members.

The endpoints of a wireless Internet will include familiar information devices carried by people (telephones, PDAs, laptop computers). There will be an increasing number of autonomous devices such as wireless sensors with specialized tasks of acquiring, transmitting, and receiving diverse types of data. A few examples are geolocation information, biomedical measurements, and surveillance pictures. Pervasive computing anticipates a proliferation of cooperating autonomous wireless terminals.

Multicasting, an increasingly popular mode of Internet information transfer, is likely to be even more attractive in a wireless Internet. In a multicast, the "end user" is a group comprising a variable population of members defined on a per-session basis. In a wireless Internet, multicasting is likely to be just as popular but, owing to the mobility of terminals and variability of transmission conditions, it will present challenges that do not arise in the wired Internet. ^[7]

Geocasting is a form of multicast that that can add to the value of a wireless Internet. ^[8] Geocasting defines a multicast group with reference to a target area. The members of the group are terminals with geographical coordinates within the target area. In addition to location, mobility states (velocity and direction) and demographics can be major factors in the definition of geocast groups. The geocast membership can be specified by the sender of information, the recipient, or by a service provider. A geocast session may consist of one or more messages that are sent to the geocast group. A message can originate with a group member or outside the group. For example, in the action "send a reminder to all students and faculty within 3 km of the campus that a seminar will begin in 30 minutes," the originator of a message defines a geocast group by location and demographic category. In the action "get information about all shoe stores that I can reach in 30 minutes," the information recipient defines a group by location and mobility. Finally, in the action "notify everyone within a radius of 10 km of a traffic jam," the service provider uses an arbitrary criterion defined by location to specify a group.

Geolocation (discussed in further detail in Section 2.4.2.1), the process of determining the geographical coordinates of an information device, is a technology that supports geocasting. The construction and maintenance of the geocast group are nontrivial tasks for mobile networks. Most studies assume that geolocation information is continuously available to mobile nodes via the Global Positioning System (GPS). While this is generally a viable assumption, the manner in which the geolocation information is acquired and disseminated has significant impact on network capacity and performance.

2.3.2 The Geography of Information ^[9]

For twenty years, the expression "anytime, anywhere" has been a cellular technology mantra. At first only a lofty goal, the combination of satellite telephones and terrestrial cellular systems have made "anytime, anywhere" a reality for telephone calls and short text messages; it is also a good description of the World Wide Web paradigm in which content seems pervasive, contained in Web pages that can be summoned to any computer in the world at the click of a mouse. Although this paradigm is appealing, the geography of signal transmission, described in Section 2.3.3, makes it difficult and expensive to achieve with wireless technology, even for the simple task of delivering telephone calls and short messages. With the added complexity of multimedia wireless Internet information and the diversity of user characteristics, "anytime, anywhere" becomes prohibitively demanding. Thus, we would do well to examine the nature of the information conveyed in a wireless Internet to determine the conditions in which ubiquitous, instantaneous coverage is essential, not merely a convenience to be weighed against its costs. Rather than impose the burden of "anytime, anywhere" on all communications in a wireless Internet, we examine the temporal, spatial, and demographic coordinates of information. Matching communication technology to information geography promises gains in efficiency and quality of a wireless Internet.

In examining the geography of information, we classify services according to where, when, and to whom the information is relevant. We represent the classification in each of these dimensions — space, time, and personal — in a range from specific to general. At one extreme we have information that is useful to only one person, at a particular time, when the person is in a particular place. For example, a message generated while you are on your way to the airport that "you are urgently requested to deal with an emergency in your home" is localized in all three dimensions. Unless you receive the message very soon and you are near home, the information is not very useful to you. Information at the other extreme is a popular music recording. It has no time localization and it is of interest to a large population of people throughout the world. Table 2.4 gives examples of information in the corners of the three-dimensional cube in which spatial relevance, temporal relevance, and personal relevance range from specific to general.

Table 2.4: Localized and General Information

Location	Time	Personal	Information Example
Specific	Specific	Specific	Emergency dispatch message
Specific	Specific	General	Traffic conditions
Specific	General	Specific	Alert us when a friend is nearby
Specific	General	General	Local maps, directories
General	Specific	Specific	Horoscope
General	Specific	General	Stock market prices
General	General	Specific	Message containing family news
General	General	General	Music recording

The first example, at the top of the table, requires "anytime, anywhere" message delivery through a network with ubiquitous coverage, while by contrast the music recording at the bottom of the table can be downloaded at a time and place that are convenient, economical, and conducive to reliable information transfer. If the recording is very popular, multicasting would make sense. Local maps and directories in the middle of the table lend themselves to geocasting by wireless information kiosks.

2.3.3 The Geography of Signal Transmission

It is well known that the signals transmitted by wireless modems are subject to a variety of transmission impairments, the most prominent of which are:

• Attenuation that depends on the distance between transmitter and receiver

• Fading that depends on the physical characteristics of the transmission environment and the motion of wireless terminals

• Additive noise in modem receivers

• Interference due to transmissions by other modems

Attenuation and fading effects are highly dependent on the locations of transmitters and receivers and interference varies with both time and the locations of the interfering transmitters and the location of the signal receiver.

Engineers have devised a vast array of modulation, reception, coding, signal processing, and network control techniques to mitigate the effects of these impairments. To use them effectively, network managers devote high levels of effort and expense to address the geography of signal transmission and the geography of users in determining the locations of base stations and access points and precisely orienting their antennas. They aim for highly reliable signal reception in the greatest possible coverage area at all times.

In spite of the effort and expense devoted to erasing the inherent time and location dependence of signal quality, the goal of "anytime, anywhere" communications remains elusive in all WLANs and new (2.5G and 3G) cellular systems. All of these technologies prescribe radio modems that can operate with a collection of modulation and coding schemes, each with its own transmission rate and immunity to impairments. They employ rate adaptation to find at any time and place the best compromise between signal quality and transmission rate. As in the example of Figure 2.2, this trade-off depends on the locations of transmitters and receivers and on network activity.

The nature of the compromise resembles that of a telephone modem built into a personal computer in that the modem operates at a bit rate matched to the characteristics of each dial-up connection. However, the effects on applications are quite different. A dial-up modem operates at one rate for the duration of a connection. By contrast, the mobility of wireless terminals and the time-varying nature of the interference will cause wireless modems to change their rates far more frequently. Managing quality of service of applications in the presence of location-dependent and time-dependent transmission rates and signal quality levels is a major challenge that remains to be addressed.

^[7]Gossain, H., de Morais Cordeiro, C., and Agrawal, D.P., Multicast: wired to wireless, IEEE Communications Magazine, 40(6)116–123, June 2002.

^[8]Navas, J.C. and Imielinski, T., Geographic addressing and routing, Proc. Mobicom '97, Budapest, Hungary, September 1997.

^[9]Goodman, D.J., The wireless Internet: promises and challenges, Computer, 33 (7), 36–41, 2000.