|Philip Robinson received his B.Sc. in Computer Science from the
University of the West Indies (UWI)
in 1996. He then worked as a software programmer until 2000, when he traveled to the UK to pursue a M.Sc. in Distributed Systems. He was then awarded with the degree of M.Sc. in Distributed Interactive Systems by Lancaster University in August 2001.
Philip has also been the co-organizer of three workshops on security in Ubiquitous and Pervasive Computing in UbiComp 2002, UbiComp 2003 and Pervasive 2004, the latter being published as a post proceedings in the Kluwer Computer Science series. The publication is entitled: “Security, Privacy and Trust in the Context of Pervasive Computing”.
The core hypothesis of my thesis is that entities in dynamic environments aim at maintaining their limited “perception of security” rather than explicitly protecting the system. If all trusted entities perceive that the system is secure, then this is the best possible security state. Furthermore, I believe that a security system may be divided into 5 hierarchical, distributable roles, which work together to form a complete security system. These roles are defined as initiator, coordinator, effector, monitor and announcer; The thesis claims that a system is most secure when all five of these management roles are assigned to and executable by trusted entities, yet allows the simplest case of all roles being carried out by a single entity.
There are two main challenges that I investigate:
Complexity of managing (specification, enforcement, update)
adaptive security and
Coordinating secure interaction between entities in dynamic
Before progressing to discuss the relevance of this thesis, I want to clarify the concept of “perception of security”: Perception is possible when an entity creates a model of the world in which it interacts. Perception is positively affected by events that confirm or enforce the validity of the model and negatively affected by events that distort the model. A perception of security is therefore an entity's model of system defined by its knowledge, assigned tasks and permissible views.
The research is being conducted in two different systems areas yet the
similar properties of dynamics and distribution have prompted this
Meeting rooms and offices outfitted with supporting technology for
ubiquitous computing. This allows new meeting attendees to
spontaneously arrive and use equipment (projector, sensors, air
conditioning) available in the room, or they may even bring their own
equipment (cameras, laptops) to further enhance the system. Consequently,
meeting attendees may also leave and take their equipment, causing the
system to spontaneously reconfigure.
Virtual organizations represent coalitions of organizations that come together
only to accomplish a business venture that could not be completed by only
one (or subset) of the organizations. Therefore they need to create
connections between their computer systems for resource sharing and
coordination of business processes. However, partners change frequently,
changing the availability of services and resources of the virtual
organization, as well as the order in which processes are executed.
following properties are used to describe both of the above system
Not possible to pre-determine actual constituents (hardware, software,
users) of systems and consequent behaviour
protection goals must be locally enforced although centrally specified
challenge of securing tasks as opposed to securing data
protection goals may invariably change based on the actual state of the
goal of the thesis is to show how dependable security in a distributed,
dynamic system can be obtained when each entity (referred to as a
controller) is given a clear description of its security assignment in the
system, having proven that it can fulfill that role. The term Authorised
Transient Control is used as a container for the research, as it
each active entity in the system must be authorized to fulfill its
security role, from which its perception of security is derived
the perception of security however frequently changes based on the
situation, and the controllers are also expected to dynamically change
(entity or role)
Control: therefore each entity of the system is assigned to control with a particular perception as reference.
The contribution of the thesis will therefore be a security management
model and coordination protocols for dynamically reconfiguring systems
with the properties identified above.
Although security has become the thematic area of my research, my general
interests are in the design, implementation and management of distributed
systems. Ubiquitous computing shares many of the features of a distributed
system, including autonomy of computational nodes, asynchronous
communication, lack of a global clock, and dynamic interchange of
resources. Therefore, many of the design, implementation and management challenges of UbiComp are also similar to those of distributed computing. Below is a list of my
general interests. I have been author and co-author in at least one publication in each of these areas.