Although the workshop of 6 and 7 March 2006 was the first in the series, it is mentioned here last, as research is the next step towards the “Internet of Things”. This network of billions of machines communicating with one another is a major theme for the evolution of information and communications over the next few decades. This development is not only the result of a technology push; technology push in this field is mirrored by market pull, with short-term objectives and tangible application prospects.

From a generic point of view, it can be concluded that the trend towards an ever larger population of integrated smart systems is irreversible because the economic value of a system of objects and devices is directly related to the fact that they are “networked”[1]. From the technological perspective, the Internet of Things will enable computing to melt invisibly into the fabric of our business, personal and social environments, supporting our economic, health, community and private lives.

The potential applications of pervasive networking are wide-ranging. To make this technology worthwhile, there is a need to address large classes of potential applications in order to better understand the various requirements (e.g., real time, quality of service) that will eventually drive the needed generic technology developments. However, application drive alone will not deliver the wide variety of solutions that have to be brought together. The list of research that needs to be undertaken is a long one that ranges across industrial sectors, beyond the classical telecommunication technology providers.

Within the competitive business conditions that prevail today, industry and other players have contributed to an endemic climate of “hype”. Economic prospects related to networking of large number of simple devices like RFID are huge. Still, the economic prospects related to more sophisticated computing devices such as sensors need further research with industrial players.

The Internet of Things is not a revolutionary concept. It fits well with the evolution of today’s networking technologies (Internet, wireless, service platforms, etc.). Even if evolutionary, a great deal of genuinely creative, innovative research is required to realise the Internet of Things. It is not simply a matter of re-engineering existing technology. Billions of connected devices are pushing current communication technologies, networks and services approaches to their limits and require new technological investigations. These cannot happen quickly and need to be tackled within a long-term perspective.

Research

In particular, research is required in the field of Internet architecture evolution, wireless system access architectures, protocols, device technologies, service-oriented architecture able to support dynamically changing environments, security and privacy. Research is required in the field of dedicated applications integrating these technologies within a complete end-to-end system.

It is important that researchers, when pursuing less constrained “blue sky” topics, should remain aware of industrial and real-world problems. Industry and academic institutions should be encouraged to keep close contact, especially where the academic research might not attract internal funding within an industrial organisation.

Research on devices

Although many examples of RFID deployment are already given, the technology is by no means at the end of its development. Costs are an important aspect: the drive for wider application depends on much lower costs. RFID costs are in the region of €0.50 for retail product tags and supply chain tracking applications, but rise dramatically beyond €25 for the more sophisticated devices used, for example, in highway toll collection. Printed circuitry is a possible approach, and a major driver in practice will be manufacturing scale and volume.

The feasibility of many applications depends on powering and power economy. Devices must have long lifetimes without the need to change batteries, yet there are many challenges in this area.

Sensor technology itself needs to evolve; in many cases, miniaturised sensors require further Research and Development (R&D). In addition, sensors have to be fitted with communication capability and also with service features to cover the most extensive use cases. These issues are clearly in the R&D domain, as the impact of these communication and service features on the devices are expected to have non-negligible cost impact.

Research on the network

On the side of the network, also a whole range of research challenges can be identified: existing data communication protocols may be inappropriate in the Internet of Things, as they rely on the transfer of relatively much metadata. Lighter protocols and lighter implementations that reduce (compress) the explicit protocol layers into a single communications module are now required.

The Internet of Things requires self-organising and self-scaling networking to arrive at mobility, economy and flexibility as well as network resilience. Self-organising networks must cope with varying requirements for physical and virtual link topologies. Service discovery and reuse by different applications has very high priority.

Furthermore, the Internet of Things must have a naming and addressing strategy by which objects can identify themselves, and can locate other objects and the communication paths to them. Because the network is heterogeneous, supporting many different devices offering different service types, a declarative interface, like Internet Description Language (IDL) or Extensible Markup Language (XML), will be needed. This will allow a device or node to describe what it presents to others.

A consistent set of middleware offering application programming interfaces, communications and other services to applications will simplify the creation of services and applications. Service approaches need to move from a “static” programmable approach towards a configurable and dynamic composition capability.

The Internet of Things must incorporate traffic and congestion management. This will sense and manage information flows, detect overflow conditions and implement resource reservation for time-critical and life-critical data flows.

Regulation

The Internet of Things is a pervasive federated network in which unregulated personal area networks and local area networks will interoperate with more traditionally regulated electronic communications services. Regulators need to carefully monitor the challenges posed by these networks, taking action as necessary to regulate for technical interoperability, consumer protection, support for competition and the appearance of opportunities for the exploitation of market power. Here, traditional infrastructure regulatory frameworks should be reviewed.

A foreseen area where there is risk of monopoly market power is that of the ownership of the data resources in name servers. These are the resources that networks must use to determine the way to reach a given person, device or resource. The implications in terms of Object Name Server (ONS) management should be evaluated. Existing regulatory issues that have been and are being tackled in existing networks, for example access, roaming and billing, may raise their heads in different guises. Also the manner of allocation and regulation of radio spectrum is a key issue for the development of the Internet of Things. Spectrum scarcity will remain an issue and flexible approaches to spectrum management have to be researched, especially for devices that will primarily be deployed in unlicensed bands.

In summary

RFID can enable the realisation of an Internet of Things. However promising many existing applications are, a lot of research is still needed before the long-term perspective can take hold. In the section above, some of the most pertinent issues are addressed, but this listing is based on workshop outcomes; a roadmap for RFID research would be needed. “Blue sky” research will be essential, though it needs to be strongly backed with industrial perspectives: companies are invited to share their problem issues with the academic world. Finally, also the regulatory and spectrum consequences of RFID are a topic for research, as was also proposed in the other workshops.