The incorporation of ontologies and semantic reasoning have far reaching implications for wireless communications and spectrum management. Early implementations using adaptive spectrum methods are already being deployed. In the future much deeper uses will be developed.
In the context of wireless communications ontology and semantic reasoning have similar roles to their use in the semantic web. The ontology layer defines the participants and variables that impact communication. It also describes the relationships between them. A semantic reasoning engine uses the ontology to analyze the current data and formulate a transmission plan to implement the policies it has been given. Thus, the work being developed for the semantic web is being applied to spectrum management.
Physical Layer Stakeholders
Early uses of policy-based radio focus on more efficient use of spectrum and with that limiting interference to acceptable levels. Even with these initial implementations priorities from different stakeholders are competing for control. Among those are:
● Network Operators
● Enterprise Network Managers
The order listed above is probably the priority each of these will get. The regulators typically have the force of law behind them. The regulatory requirements cannot be violated by any other set of priorities. The other stakeholders may pursue their priorities but must always be compliant with regulatory restrictions.
The network operators seek to run the most efficient network possible. They will not allow individual businesses or users to do things that will degrade their network performance. However, within the constraints of regulator and network operator policies, individual businesses and users have flexibility to define their own priorities.
To implement policy-based radio the reasoning is moved into the network itself. The network and its participating devices sense the spectrum and develop a transmission plan. This plan is highly dynamic. A transmission plan must be developed that satisfies the policies of all stakeholders, the regulator, network operator, enterprise network manager and user. Then the transmission plan must be executed while the spectrum data is still relevant! Accomplishing this requires high speed and highly sophisticated analysis of the spectrum data.
Once the reasoning is moved into the wireless network far reaching possibilities become available. The reasoning can use the spectrum data but could go further. The network can also reason about the source and content of the transmission. A commonly used example is “John gets Gold service”:
John is the new CEO of the company. The IT department gets the order to setup his devices and services, but someone forgot to check the box to give John Gold service. All executives at this company get priority treatment called Gold service. As a result, IT sets up John for regular service. However, the network has access to the HR database and can ‘see’ that John is the new CEO. Including that information in its reasoning it concludes that John should be setup for Gold service, rather than regular service. The network, not a person, automatically corrects the problem and John has Gold service from his first day on the job.
In this example data available through a semantic web is utilized to help determine the communication service level provided. The convergence of these area, which are now largely occurring in separate fields, results in impressive new possibilities.
As networks gain the ability to reason about the data, far reaching possibilities become viable. An example in the business realm might be:
The company board wants to improve customer satisfaction. They request plans for achieving that goal and select one, ordering that it be implemented ASAP. The network does a semantically based analysis of the plan and concludes that its implementation will cost $100M more for new servers and IT staff than the current budget allows. A message is sent to the board, while it is still in session that with the order to implement the plan, the board needs to add $100M to the IT budget or there will be problems.
Application of ontologically based networks and semantic reasoning has profound implications for mission planning:
A mission commander orders a mission to go to a remote location, achieve a certain objective and return in three days. He sends an order for that mission. The network reasons through that order and sends the mission commander a message. In that message the commander is reminded that there are two other missions already in the field and all are now scheduled to return on the same day. However, there are not enough operational helicopters to support all three missions. There is a shortage of a critical part and more helicopters cannot be made operational for another week. Because the network understands the factors involved and the relationships between mission orders, operational helicopters, and spare parts logistics, it can inform the commander that his order cannot be implemented. Modified orders must be developed.
Implementing Ontological and Epistemological Logic
Implementing ontologically based thinking in future networks will require inclusion of epistemological concerns. Ontology is the study of the nature of things and the relationship between objects in a domain. Epistemology concerns itself with the accuracy and correctness of information. As applied to spectrum management, ontological reasoning understands the connection between spectrum data and a specific device’s transmission plan. Epistemology deals with the reliability of the spectrum sensors and their data. As the old saying goes, “garbage in, garbage out”. If the spectrum sensors are too limited or if the sensor data is not accurate, then the system will not be able to develop a good transmission plan.
Future networks will need to need to safeguard the quality of the data, the policies being implemented and the process to act to optimally apply those policies. They must then be sure that the relationships and interactions between various objects and factors are accurate. Both the ontology and epistemology must be reliable if the reasoning is to be correct and the transmission plan developed is to be accurate.
Implementing these new networks will require development of new programming languages with built in support for ontological relationships and relative levels of policy authority.
Ideally, these new languages will be able to take in requirements written in typical regulatory language and implement those requirements directly, without the need to convert them to some machine-readable form. Every translation creates the potential for translation mistakes. In addition, translation takes time. Ideally policies can be applied in their original format. Regulators are often attorneys and write like attorneys. Business managers also have their own writing style and can be expected to create their policies in that style. The ability to implement original content improves speed and also eliminates the potential for translation errors.
These new networks will present great potential to serve positive social goals but equally could be turned to do great harm. The security concerns are substantial. Any powerful technological development can be used to benefit but has the potential for malicious use. To protect against malicious uses, new security methods will be required. Further, the security design must be integrated into the design, not just added on at the end.
Future wireless networks are being developed with fundamentally different capabilities. The use of ontologies and semantic reasoning will enable networks to go much further than just following a static set of rules. These networks will be able to use spectrum adaptively, increasing the spectrum efficiency by orders of magnitude. They will be able to go far beyond managing static transmission plans. They will incorporate reasoning as a native part of network function.
You may also enjoy these stories by Stephen Berger:
Radio is changing from hardware to software: new policies needed! Perhaps surprisingly, theology may play a role in determining how radio will be used.
We are running out of radio spectrum but AI might help.