MoN9: Ninth Mathematics of Networks meeting, 18th June, St Andrews University

Rashid Mehmood (Swansea) - Verifying feature interactions in intra- and inter-vehicular networked systems

This is joint work with Atif Alvi, LUMS, Pakistan and David Greaves, University of Cambridge, UK.

Intra-vehicular networks are no more simple and static. They are increasingly complex because they comprise increasing number of intelligent sensors, electronic control units (ECUs), and comfort and entertainment appliances. They are increasingly becoming dynamic because the network architecture is evolving to allow runtime binding of additional application, services, and networks. The drivers for these developments include the changing landscape of software and hardware, miniaturisation of electronic devices, demand for latest in-vehicle consumer electronics, and gaps between automotive and consumer electronics lifecycles.

Inter-vehicular networks are driven by the Intelligent Transportation Systems agenda with the aim of providing enhanced safety, transport efficiency, comfort and facilities. Rapid developments in computing and wireless communication technologies have enabled the possibilities for vehicles travelling at high speeds to form ad hoc networks, on the fly, in order to coordinate and collaborate for shared benefits. For examples, vehicles can coordinate on junctions, roundabouts and slip roads to improve traffic flow and reduce accident possibilities.

Inter-vehicular networks form a complex pervasive computing environment where feature interaction takes place between independent and intelligent vehicle entities or vehicle sub-networks. Similarly, intra-vehicular networks form another layer or hierarchy of pervasive environment where feature interaction takes place between sensors, ECUs, consumer electronic appliances. We can think of a feature as a unit of functionality - a facet of behaviour - of a system. This means that the functionality of different entities in a vehicular network represent features and have the potential to interact in a conflicting manner. It is a major research challenge to predict, identify and resolve feature interactions, as these may be damaging to various system dimensions including privacy, security, functionality, and facility.

We report on our work using feature interaction methodologies to design and verify vehicular networks, both intra- and inter-vehicular, against the user defined policies for privacy, security, functionality and facility. Our approach is a hierarchical one where, to address scalability, we divide the system in multiple hierarchies and verify the whole system by approaching individual entities and subsystems, one at a time. The hierarchical design ensures that all the network entities (microcontrollers, sensors, actuators, ECUs, vehicle) are formally verified to cause no feature interaction with other entities and to obey some standing rules defined by the designers of the system. An ontology substrate is used to hold the attributes of the system to perform various types of reasoning over them: consistency checking, classification, and enforcing higher abstraction rules in first order logic that are authored by the designers and by the passengers. These higher abstraction rules form the system level.

We will elaborate through examples as to how our methods could be used to predict, identify and resolve system requirements such as privacy, security, functionality and facility. The work is ongoing and is being extended in terms of other pervasive environment and the improvements in methodology.

Return to previous page

Contact: Richard G. Clegg (richard@richardclegg.org) or Keith Briggs (mailto:keith.briggs_at_bt_dot_com)