“…The problem considered by the authors in [6] is then to compute the largest possible collaboration between a collection of systems which still remains secure, and the authors show that this problem is in fact NP-complete, thus showing that it is impractical to compute this largest set for a large array of distributed systems. Shehab, Bertino, and Ghafoor [18] expand on a variant of this model; their work presents a framework similar to that modeled by Gong and Qian, [6] [7], but in the absence of any trusted third-party mediator having a global view of access control policies and thus avoiding associated bottleneck issues associated with such a mediator.…”