Waveform Agility and the Next Generation Internet
Three disruptive trends are in play, and together will likely have a signficant impact on the next generation Internet architecture.
First, recent years have seen the emergence of a number of wireless communications technologies such as adaptive beamforming, MIMO, Ultra-wideband (UWB), and spectrum-agile communications. Each of these technologies has its own unique profile in the kind of links created between communicating nodes. For example, MIMO works very well in a multipath-rich environment yielding high data rates, but introduces processing latency; UWB is high capacity and interference-friendly but short range. This diversity is only increasing.
Second, software radios are quickly emerging as the platform of choice for future wireless communications systems, from commercial  to open source  to military . In a software radio, many signal processing functions such as modulation, coding, and spreading are done in software. This enables agile radios - a software radio that uses its flexibility to dynamically change waveform characteristics and behavior in response to instruction and even cognitive radios - radios that know about their capabilities, internal state and the spectrum environment to opportunistically communicate (without explicit instruction) using the best combination of
transceiver parameters . The trend is clearly toward an increasing flexibility, agility and autonomy in the selection of wireless communication waveforms.
Third is the emergence of Mobile Ad Hoc Networks (MANETs) in which network nodes, possibly mobile, can self-organize and communicate wirelessly over multiple hops. With MANETs, it is possible to quickly and opportunistically extend the reach of the Internet to remote areas and mobile nodes. It is likely that the next generation Internet will be based on a hybrid wired/wireless infrastructure with access points, rooftop nodes, and other mobile nodes self-organizing to form a mesh network. The rapid increase in the number of standards bodies interested in mesh networking is a testimony to its importance – currently IEEE 802.11s, 802.16a, IETF, IEEE 802.15.5 and IEEE 802.20 are working on various aspects of mesh networking .
The confluence of these three trends has significant implications for the future of the Internet. The combination of software/cognitive radios and diversity of emerging waveforms will likely result in a single node being able to engage a number of waveforms, with uparalleled agility between and within these waveforms. In other words, we have a node that can pretty much create any kind of communication link it wants, with whatever combination of capacity, error-rate etc. it needs (subject to what is physically achievable, of course). And in the next generation Internet, which includes mesh networks as an integral part, such links will be part of the topology.
How does this impact the next generation Internet architecture? Waveform agility enables the concept of a definable link, as opposed to the conventional notion of simple, fixed links. In other words, a link is no longer just an input to the topology and the algorithms that operate on the topology (e.g. OSPF). It is a variable, a parameter that can be controlled as necessary by the protocols. An architecture should not only tolerate the highly dynamic nature of links, but should also be able to tailor the link characteristics in a cross-layer fashion to create the kind of topology that an application requires. In other words, topology control – control of which nodes should be able to communicate as well as how they should communicate – should be an integral part of the architecture.
Several key requirements for the next generation Internet architecture arise:
- Unlike the current layered architecture, which encourages abstraction and “hiding” of one layer’s details from the higher, all control mechanisms should be cognizant of and be able to harness the physical layer flexibility. While cross-layer approaches are promising, we should go beyond that and investigate alternate layering hierarchies.
- Routing protocols should not only be able to accommodate mobility of infrastructure elements, but also be able to accommodate and control the diversity and agility of link characteristics. In particular, it should be able to vary the communication waveforms and parameters to create topologies with the “right” combination of capacity, robustness, error characteristics etc. It must be able to set up links “on demand”, for instance using “message ferries ”
- Transport protocols must be cognizant of the diversity and agility of the underlying infrastructure and should incorporate feedback control and reliability mechanisms that are adaptive to a wide range of media characteristics. Further, they must be delay tolerant and disruption tolerant, and work under the “eventual connectivity” model, that is, networks that are disconnected at any particular instant but that have an end-to-end path over a period of time .
- Cognitive radios and spectrum agility have new implications with regard to spectrum policy. Network control mechanisms should be able to negotiate and select spectrum opportunities in accordance with policy. This requires an architecture where spectrum and other policies are machine-encoded and nodes reason about policies in the context of their own capabilities and sensing of the environment .
One of the challenges in accommodating wireless and mobile nodes in the next generation Internet is the performance difference between wireline and multihop wireless networks. This is especially serious if MANETs or mesh networks are woven into the Internet infrastructure not just at the edge (stub networks) but also serve as transit (core routers) . MANETs continue to lag behind their wireline counterparts in terms of latency, capacity and robustness. Research is needed to elevate MANETs to a performance plane on par with wireline networks, especially in the area of medium access control and cooperative diversity in a multihop wireless context .
In sum, the next generation Internet architecture needs to allow for a high degree of flexibility and adaptivity in order to accommodate the diversity and agility of waveforms in a hybrid wired/wireless infrastructure context. Significant changes to the routing and transport protocols are required, and policy-driven control is key to accommodating cognitive radios. These changes may well require a complete re-think of the layering hierarchy as we know and use it today. Experimental testbeds in support of research in this area should target flexible and reconfigurable switching nodes with flexible link-layer interfaces that can be configured for a range of wireless link capabilities.
- J. Mitola III, G.Q. Maguire, "Cognitive Radio: Making Software Radios More Personal," IEEE Personal Communications, August 1999.
- R. Bruno, M. Conti, E. Gregori, "Mesh Networks: Commodity Multihop Ad Hoc Networks," IEEE Communications Magazine, March 2005
- W. Zhao, M. Ammar, E. Zegura, "A Message Ferrying Approach for Data Delivery in Sparse Mobile Ad Hoc Networks," Proc. ACM MobiHoc, May 2004, Roppongi, Japan
- V. Cerf, et al, "Delay Tolerant Network Architecture," IRTF Draft, draft-irtf-dtnrg-arch-03.txt, July 2005
- R. Ramanathan, "Challenges: A Radically New Architecture for Next Generation Mobile Ad Hoc Networks," Proc. ACM Mobicom 2005, Cologne, Germany, Aug. 2005 (to appear).