There was general agreement among Workshop participants that significant research progress in wireless and mobile communication systems and networks would require an interdisciplinary approach that cuts across all layers in the networking hierarchy. This is in distinction to the majority of existing work in this area where, typically, communications researchers concentrate almost exclusively on lower-layer issues in relative isolation from the work of networking researchers with a major focus on higher-layer issues. The unprecedented complexity associated with the design and implementation of efficient wireless and mobile communication systems and networks will require much closer cooperation and coordination between the research efforts of these two groups. Every effort should be made to encourage such cross-cutting research approaches.
A number of potential cross-cutting research efforts were identified, several of which are discussed below. These examples were intended to provide selected illustrations of representative cross-cutting research efforts and are in no way to be considered exhaustive of the potential for such work. Specific examples included the following:
The end-to-end propagation delay in global networks is already large enough to severely stress allowable budgets for network induced queuing, processing and transmission delays. This is particularly the case for delay intolerant traffic such as multimedia services. Furthermore, channel variability on wireless subnetworks generally require event- driven adaptive protocols. Since adaptive protocols typically introduce additional delay, this will provide further stress on the overall delay budget. Satisfaction of this overall delay budget will require careful allocation of delay across protocol layers which in turn will require cross-cutting research efforts between communications and networking/protocol researchers.
A suggested possible approach to providing this delay allocation across protocol layers is by exploiting basic performance/delay tradeoffs. More specifically, it was felt that there exists fundamental performance/delay tradeoffs at each networking layer, although these tradeoffs are perhaps best understood at the lower layers. The key to delay allocation across protocol layers is then to develop a better understanding of these basic tradeoffs and how they can be used in appropriate design methodologies to meet overall delay budgets while satisfying imposed QoS requirements. This is an example of a representative cross- cutting research topic that provides a natural bridging across protocol layers.
The increasing heterogeneity of future communication systems and networks, both in terms of network architectures and terminal capabilities, poses a number of problems in their efficient design, implementation and operation. This is particularly the case in the internetworking of wireless and mobile subnetworks to wireline backbone networking infrastructure and is most evident in the transport of evolving services such as multimedia.
The heterogeneity can perhaps best be addressed through use of some form of scalable delivery approach where the source material is represented in terms of hierarchical layers with each layer handled differently by the network. This generally implies layer-specific QoS requirements from the network with some form of synchronization constraint on the delivery of different source layers. The layer-specific QoS network requirements will depend very much on the application. For example, it can be expected to be very much different for MPEG-X encoded video for entertainment purposes than for H.26X encoded video for videotelephony applications.
While much work has been done in the development of scalable source representations, little work has been done in addressing the network transport issues and, more specifically, developing efficient overall transport approaches with appropriate layer-specific QoS support. These are issues which span multiple networking protocol layers and require a cross-cutting research approach. Innovative research which approach this problem across multiple network protocol layers are expected to have significant potential and are specifically encouraged.
Encryption for data confidentiality is generally considered exclusively a higher-layer issue, typically implemented at the presentation layer. As indicated previously in this report, security issues for wireless and mobile communication systems and networks are much broader and more comprehensive than merely data confidentiality. An appropriate approach to the overall security issue for such networks then is not expected to be localized to a single network protocol layer but to span multiple protocol layers. This is an other area where effective cross-cutting research can have considerable impact.
Indeed, even for data confidentiality the localization of encryption to a single protocol layer needs to be reconsidered for wireless and mobile networks. The provisioning of encryption as a presentation layer function is generally based on the premise of reliable message delivery provided by lower layers. This premise is somewhat questionable for representative wireless networks and particularly so for delay intolerant traffic such as multimedia. More specifically, basic delay/performance tradeoff criterion would suggest that delay intolerance implies unreliable delivery. As a result, encryption/decryption techniques that require reliable delivery are of questionable value for delay intolerant services on wireless and mobile networks and research is needed to develop new approaches that remove this requirement. It is expected that these new techniques cannot be localized to a single network protocol layer but are distributed across multiple layers. Furthermore, the data confidentiality aspect of network security needs to be integrated with the other aspects, such as validation and authentication, which already distributed across multiple layers.
Another area where innovative cross-cutting research is expected to have considerable impact is in providing mobility management for wireless and mobile communication systems and networks. User mobility introduces new connection management issues which requires frequent updating and realignment of user files, registration databases and routing tables. All of this introduces considerable transmission overhead and associated latency not generally experienced on fixed wireline networks.
The mobility management issues can generally be addressed using an appropriate self-organizing, distributed control network. How the mobility management functions are to be distributed across network protocol layers is presently unclear and requires considerable research to define, investigate and characterize promising approaches. It is clear, however, that this coordination will span multiple layers right down through the physical layer. For example, some form of user location/tracking will probably be required to reduce the latency associated with updating and caching files in the face of user mobility. This is a function that probably already is provided in the physical layer and can be effectively used at higher layers as well as part of an overall mobility management approach.