Dr. Matta's contributions span the design and analysis of transport and routing protocols, on top of multi-hop wired and wireless networks. These protocols naturally implement constrained optimization solutions in a distributed way. Dr. Matta's focus has been on the resiliency and safety aspects in the presence of non-adversarial as well as adversarial changes.
Architecturally, Dr. Matta's approach advocates in-network storage and management, and feedback (measurement)-based traffic control. This is exemplified in the following work with his collaborators:
The techniques used to model and evaluate the developed designs come from several theories, including probability theory, queuing theory, dynamic flow theory, network calculus, numerical analysis, control theory, and information theory. Recently, together with his collaborators, Dr. Matta developed the TRAFFIC framework (inspired by the iBench philosophy) which advocates the use of programming language (type) theory that leverages all aforementioned (lower-level) theories to define (higher-level) "types" of input/output interfaces for various network elements, together with relevant typing hierarchies. This enables safe and efficient large-scale (Internet) programming in which users can compose various end-to-end services as long as interfaces are "compatible".
Dr. Matta and his collaborators also examined the resilience of feedback-based control protocols to adversarial users who launch stealthy (low-rate) attacks. This work coined the term RoQ (Reduction of Quality) attacks to contrast these attacks to traditional sustained (high-rate) attacks. Systems designed for only non-adversarial workload are shown to be quite vulnerable to RoQ attacks, for example, the vulnerabilities of transport protocols, admission controllers, and load balancers have been assessed.