2.4 Research Initiatives

2.4 Research Initiatives

The industry trends described in Section 2.2 follow evolutionary paths within the framework of established cellular and local area networks. In doing so, they fail to address directly the essential characteristics of a wireless Internet as represented by the geography of users, the geography of information, and the geography of signal transmission described in Section 2.3.3. To fill the gap, the research community is exploring novel network architectures supported by an IP-based core network. The network architectures include proximity-based communications, ad hoc networking, and hybrids incorporating these with cellular networks, WLANs, and other approaches. The IP core network will require a new service support sublayer between the transport layer and the applications layer. It will make use of geolocation information to facilitate network control and optimize the use of network resources. This section describes examples of work in progress on adaptive network architecture and on the core network.

2.4.1 Adaptive Network Architectures

A wireless Internet presents new possibilities of adaptive networking solutions. Instead of simply cutting the wires in the last mile and viewing air as the hostile medium of transmission whose shortcomings need to be combated, the absence of wires can provide novel means of disseminating control and user information by taking advantage of mobility. The client/server model that governs the cellular approach, where all radio resource allocation is determined centrally, attempts to erase the effects of mobility rather than take advantage of it. As mentioned in Section 2.3.1, the diversity of users accessing a wireless Internet will proliferate with the future deployment of autonomous devices that collect, measure, process, query, and relay information. The growth in pervasive computing devices will make it impractical for fixed access points to provide centralized mobility management and ensure bandwidth efficiency. Therefore, in the confines of the client/server model of the cellular architecture, mobile users would experience limited quality of service and limited data access.

The radio links of a wireless Internet are at the perimeter of a complex information network. The interface between the core Internet and the radio links of a wireless Internet comprise a radio access network, which needs to respond to the changing wireless landscape. Radio access networks for cellular and WLAN radio links differ significantly. Cellular access networks consist of a sophisticated infrastructure linking base stations, routers, servers, and databases. WLAN access networks come in many varieties. They can have a substantial infrastructure or none at all, relying on ad hoc connections between WLAN modems for network control. The networks can have a hierarchical or peer-to-peer topology. However, no radio access network has yet emerged as the clear winner. The next-generation wireless network will need to be a network of networks where the boundaries between different modes of radio access are transparent to the user. The quest for this capability is reflected in the research-and-development efforts toward WLAN/cellular coordination discussed in Section 2.2.3. More radical examples of seamless transition between modes of access in real-time are emerging in the research community.

Future radio access networks will need to promote efficient use of electromagnetic resources by all transceivers, mobile and fixed. This mindset immediately points to the flexibility offered by proximity-based and peer-to-peer communications augmenting the conventional infrastructure networking. Such flexibility would present adaptation capability to overall spatial and temporal variation of traffic, as well as the mobility states of user groups and individual users. Further, mobility of nodes can now be viewed as an advantage in information dissemination, routing, and cooperation for improved quality of service. In an adaptive network of mobile nodes, each mobile device enriches the web of communication by contributing to the network density. Data can move from car to car, among people passing each other in the streets, in the hallways of an office building, in a park, or in an airport. Military communications development efforts have for some time taken these concepts into account. Proximity-based communications will promote the efficiency and resilience of a wireless Internet.

Solutions that take advantage of proximity and peer-to-peer cooperative communications include Infostations, multihop systems that extend the range of fixed wireless systems, ad hoc networks of various hierarchy levels, and hybrid systems. The common thread in this seemingly diverse set of architectures is the motivation to adapt to the geography of users, information, and signal transmission in a locally optimal manner. This section is a survey of exemplary new architectures.

2.4.1.1 Proximity-Based Systems

An Infostation is an example of a proximity-based system that takes advantage of user mobility; ^[10] it provides wireless information services to users located in or traversing a limited coverage area. As people with notebook computers or PDAs pass by an Infostation, they receive useful information, with little or no human interaction. This could be information that is most relevant near the Infostation, such as local maps, restaurant listings, or information about courses at a university; or it could be information of general interest, such as news articles or music.

As shown in Figure 2.3, an Infostation consists of a radio transceiver (such as a WLAN access point) that provides high bit rate, low-cost, low-power network connections to portable terminals in a restricted coverage area, along with computer hardware and software that caches relevant data and schedules transmissions. Because a subscriber to an Infostation service may spend a short time in the service area of each Infostation, the information transfer should be organized in advance and should take place at the speed of electronic processes rather than the speed of human-computer interactions. A network of Infostations consists of several isolated coverage areas separated by large gaps. The cellular network serving the region containing all the Infostations can enhance the operation of the Infostation network by observing the changing locations of users. The Infostation system can use this location knowledge to move information needed by a user to an Infostation before the user arrives. ^[11] The information can be quickly downloaded to an information terminal in the short time that the user is in the Infostation coverage area.

Figure 2.3: Infostation system elements.

The Infostation paradigm is motivated by our earlier observation that no single "one size fits all" technology is suitable for all wireless information services, and WLANs and cellular networks both will be prominent in a wireless Internet. The best way to deliver information depends on various facets of the geography of the information, including spatial and temporal aspects, as well as characteristics of the users. For many types of information, "many-time, many-where" coverage offered by Infostations is sufficient to serve a mobile population. For services such as e-mail, voice mail, maps, restaurant locations, and many others, there is no penalty incurred by waiting until a terminal arrives in range of an Infostation, provided the user is sufficiently mobile. If information delivery becomes urgent, the cellular network is available to deliver it, albeit at lower bit rates and higher cost in fees and power dissipation.

2.4.1.2 Cooperative Communications

A large body of research in progress anticipates that devices will cooperate with one another to deliver information. The cooperation can occur at different protocol layers. At the physical layer, one device can provide diversity transmission and reception for another one. At the application layer, a user can receive a Web page from the cache of a nearby user and avoid the need to communicate with the Web server where the page originated. Multihopping is an example of cooperation at the network layer.

Cooperation naturally relies on proximity of network devices that can assist one another. Mobility enhances cooperation by increasing the probability that a device will be able to receive assistance from other devices. ^[12] The following paragraphs refer to work in progress on cooperative systems.

The large body of current research on ad hoc networks anticipates cooperative communications in many forms. Demon Networks (http://www.winlab.rut-gers.edu/~crose) envision an ad hoc local area network that takes advantage of mobility in order to route information. The network is not necessarily fully connected at any given time. Therefore, changes in network topology are essential for packet delivery rather than a complication to be overcome. Mobile stations can keep packets they receive and each packet to be delivered will almost certainly have many copies in the system at a given instant. The dissemination of information in this case resembles an epidemic, in which useful information is the contagious disease. The destination can be a single node or a multicast group. Each node is responsible for managing its memory allocation by making timely decisions regarding the deletion of packets it carries and disseminates.

The Terminode project applies this idea in a metropolitan area. ^[13] It uses mobility of users to disseminate information throughout a city. Each Terminode contains a map of the city and uses the map to make routing decisions.

Other forms of cooperation in ad hoc networks have been formulated as power combining and cooperative coding where diversity is exploited for purposes of maximizing bandwidth efficiency, extending coverage, network lifetime, and battery life. ^[14], ^[15]

2.4.1.3 Hybrid Architectures

In addition to the limitations of cellular infrastructure in terms of quality of service and data rates, there are situations (for example shadowing and equipment failure) in which no communication infrastructure is available to terminals. The 7DS system (seven degrees of separation) is motivated by the limitations of dependence on a network infrastructure. ^[16] In 7DS, mobile and stationary terminals cooperate to share information, help maintain connectivity to the network, and relay messages for one another. They can serve as ad hoc gateways into the Internet.

The 7DS architecture allows peer nodes to communicate via a WLAN, forming a flat ad hoc network, where some nodes have connectivity to the Internet. The connection to the infrastructure can be achieved by any access mode, such as Infostations, cellular base stations, or WLAN access points. For purposes of information sharing, peers query, discover and disseminate information. When the network connection sharing is enabled, the system allows a host to act as an application-based gateway and share its connection to the Internet. For message relaying, hosts that do have access to the Internet forward messages on behalf of other hosts. The system is an example of adaptive networking that adjusts its routing, cooperation, and power control based on the availability of energy and bandwidth. Furthermore, 7DS inherently exploits host mobility. Currently, the peer-to-peer portion of the network is implemented in a WLAN environment.

The hybrid Cellular Ad Hoc Augmented Network (CAHAN) has been developed with the above influences. ^[17] The goal of CAHAN is to make the best use of the cellular infrastructure where the centralized control and the fixed reference points provided by base stations are advantageous, and to incorporate peer-to-peer communications to optimize radio resource allocation, resilience, and power consumption. Figure 2.4 depicts an exemplary snapshot of CAHAN.

Figure 2.4: An exemplary snapshot of a CAHAN.

2.4.2 The IP-Based Core Network

It is widely accepted that future networks will converge to an IP-based core at the transport layer. In this event, an additional mobility layer between the transport and application layers is needed to ensure locally optimal wireless access to Internet services and applications. This layer will provide the intelligence for location and context awareness, media conversion, scaling, and seamless transition between modes of access in a manner that is transparent to the application or service. This layer introduces the true spirit of pervasive networking, where distributed computing and wireless access combine to make the network virtually disappear in the eyes of the user. In Yumiba et al., ^[18] the intelligent mobility support layer is referred to as service support middleware consisting of several functions grouped in two sublayers. The service support sublayer performs location management, media conversion, and user profile management. The network management sublayer performs billing, security, and QoS provisioning.

The interaction between the two sublayers and the individual function blocks in these sublayers are under investigation by several working groups. The implementation of mobility management functions in this new context is the subject of intensive, ongoing research all over the world. The evolution of GSM systems into the IP-based future cellular network is discussed in Park. ^[19] Li et al. ^[20] present an architecture that supports delivery of advanced services through WLAN access points.

The service support sublayer needs input from the radio access network on the geolocation and mobility state of the user, in addition to the mode of access. Combined with prior user information maintained in user profiles, the coordinates, speed, and direction of a user as reported by the radio access network will be translated into immediate requirements for serving a user, as well as predictions of future needs. The flow of information between the radio access network and the service support sublayer will ensure appropriate service delivery on demand or in a proactive fashion. The enablers of such pervasive networking are new technologies in geolocation, prefetching, caching, and radio resource allocation.

2.4.2.1 Geolocation

Geolocation is the term coined for determining the geographic coordinates of a mobile node. Many methods of using the radio access network with or without the aid of specialized mobile terminals have been proposed in recent years primarily with the objective of locating emergency callers. ^[21] Geolocation also tracks terminals by measuring speed and direction. All geolocation mechanisms consist of acquisition, computation, and storage of measurements or computed coordinates for averaging or tracking. The distribution of these functions is a design decision reflecting the distribution of functions between mobile terminals and fixed network elements.

Most studies of location-based services assume that geolocation information is continuously available to mobile nodes via the Global Positioning System (GPS); ^[22] while this is generally a viable assumption, the manner in which the geolocation information is utilized has significant impact on network capacity and performance. We characterize the impact of the geolocation method by its error region size. The error region is the area around the actual position of the mobile that the computed geolocation will lie in with a given high probability. The optimization of the geolocation and update intervals will mostly depend on the mobility patterns of the mobile nodes, as well as the target area size.

As an enabler of wireless Internet, a geolocation method of choice should be dictated by the requirements of the location-aware application or service. Furthermore, the signaling flow for geolocation will be different in different network architectures. Measurement, computation, and storage functions can be performed by different network components, in a hierarchy or in a cooperative, peer-to-peer fashion. The geolocation method, along with the mode of communication, should be optimized locally, so that it will work in harmony with the adaptive network architectures. The role of geolocation with respect to the geographical framework in Section 2.3 is summarized as follows: Geolocation enables the adaptation to the geography of information and users by facilitating location-aware services and applications. It helps the radio access network cater to the geography of users and the geography of signal transmission.

Facilitation of location-aware services and applications prompts signaling between the radio access network and the service support sublayer. Media conversion, content, and location management components need to interpret the position, speed, and direction of the mobile obtained from the radio access network by cross-referencing these with target area and subscriber profile information, which is maintained in the service support sublayer. The amount of signaling between the service support sublayer and the radio access network needs to be optimized. Mobility modeling and trajectory prediction methods are found to be helpful in assigning the maximum geolocation update interval subject to the quality of service requirements of the particular application or service. ^[23]

Along with mobility modeling and trajectory prediction, geolocation and tracking mechanisms make it possible for the network to prefetch and cache information proactively. An example of file prefetching in a drive-through Infostation system is given in Iacono and Rose. ^[24] Other studies consider the file prefetching in base stations of the cellular infrastructure. ^[25]

2.4.2.2 Resource Management

The ad hoc networks investigated in Section 2.4.1 raise new issues related to management of radio resources and management of battery energy in terminals. In these networks, terminals sometimes function as endpoints of communication links and other times as relays, receiving and forwarding packets moving to and from other terminals. Each terminal therefore will use some of its energy for sending and receiving its own data and another portion of its energy assisting other terminals. Routing algorithms have a strong effect on overall energy consumption in a network and in individual terminals. They influence the proportion of energy each terminal expends for itself relative to energy used to assist other terminals. Research on energy-efficient routing in ad hoc networks considers total energy consumption in transmitting a message as well as average energy consumed by the terminals that participate in the transmission. Most of this work examines a stationary network with terminals in random positions. In this situation, there is considerable variation from terminal to terminal in the proportion of energy used for the tasks of relaying and communicating. By contrast, a recent study shows that when terminals are mobile, the variation is considerably diminished. ^[26] However, truly adaptive techniques that will optimize routing decisions based on the instantaneous remaining battery power of each node need to be devised.

Game theory appears to be a promising approach to the management of battery energy in the terminals of a network of cooperating nodes. A game theory formulation defines a utility function for each terminal and an overall ("social") utility function for the network. The utility function relates to the amount of data sent and received by the terminal and the energy consumed. Each terminal adopts a strategy for maximizing its utility. Because many terminals in a network share radio resources, the strategy adopted by one terminal affects the utility obtained by the others. This situation is similar to the one addressed in research on power control in cellular systems, ^[27] in which game theory strategies led to the design of efficient algorithms for power control for cellular data. In applying game theory to resource management in networks of cooperating nodes, a major issue is the nature of the cooperation that will promote effective distribution of radio resources and fair expenditure of battery energy across the terminals in a network. It is clear that a completely noncooperative game produces suboptimum results. In the studies of cellular data systems, the base station can coordinate the cooperation among terminals. ^[28] On the other hand, the terminals in the ad hoc networks under investigation are not in communication with a single coordinating device. Therefore, the cooperation must be distributed among the terminals in the network.

^[10]Goodman, D.J. et al., Infostations: a new system model for data and messaging services, Proc. IEEE Vehicular Technology Conference, 969–973, 1997.

^[11]Iacono, A.L. and Rose, C., Bounds on file delivery delay in an Infostation system, Proc. IEEE Vehicular Technology Conference, 2295–2299, 2000.

^[12]Grossglauser, M. and Tse, D., Mobility increases the capacity of ad hoc wireless networks, IEEE/ACM Transactions on Networking, 10(4), 477–486, 2002.

^[13]Blazevic, L., Giordano, S., and Le Boudec, J.-Y., Self organized routing in wide area mobile ad hoc networks, Proc. IEEE Globecom, 5, 2814–2818, 2001.

^[14]Catovic, A. and Tekinay, S., A New Approach to Minimum Energy Routing for Next Generation Multihop Wireless Networks, J. Communications and Networks, 4(4), 351–362, 2002.

^[15]http://eeweb.poly.edu/~elza/Publications.htm.

^[16]http://www.cs.columbia.edu/~maria/project/.

^[17]Tekinay, S., Adaptive networks for next generation wireless communications: the growing role of peer-to-peer communications, in Wireless Communications and Networking, Sunay, O., Ed., Kluwer, Dordrecht, Netherlands, in press.

^[18]Yumiba, H., Imai, K., and Yabusaki, M., IP-based IMT network platform, IEEE Pers. Commun., 8 (6), 18, 2001.

^[19]Park, J.-H., Wireless Internet access for mobile subscribers based on the GPRS/UMTS network, IEEE Communications Magazine, 40(4), 38–49, 2002.

^[20]Li, J. et al., Public access mobility LAN: extending the wireless Internet into the LAN environment, IEEE Wireless Commun., 9(3), 22–30, 2002.

^[21]S. Tekinay, Guest Ed., Wireless geolocation systems and services, IEEE Communications Magazine, 36, 36(4), 28, 1998.

^[22]Sarikaya, B., Ed., Geographic Location in the Internet, Kluwer, Dordrecht, 2002.

^[23]Choi, W.-J. and Tekinay, S., Mobility modeling and management for next generation wireless networks, Proc. Symp. on Wireless Personal Multimedia Communications 2001, Aalborg, Denmark, 2001.

^[24]Iacono, A.L. and Rose, C., Infostations: a new perspective on wireless data networks, in Next Generation Wireless Networks, Tekinay, S., Ed., Kluwer, Dordrecht, Netherlands, 2001.

^[25]Kobayashi, H., Yu, S.-Z., and Mark, B.L., An integrated mobility and traffic model for resource allocation in wireless networks, Proc. Workshop on Wireless Mobile Multimedia, Boston, 3–63, 2000.

^[26]Catovic, A. and Tekinay, S., A New Approach to Minimum Energy Routing for Next Generation Multihop Wireless Networks, J. Communications and Networks, 4(4), 351–362, 2002.

^[27]Saraydar, C.U., Mandayam, N.B., and Goodman, D.J., Efficient power control via pricing in wireless data networks, IEEE Trans. Commun., 50(2), 291–303, 2002.

^[28]Goodman, D.J. and Mandayam, N.B., Network Assisted Power Control for Wireless Data, Mobile Networks and Applications, 6(5), 409–418, 2001.