1.1. Autonomic Trust Management
Autonomic Trust management is the technology to automatically evaluate, establish, maintain, and reevaluate, reestablish and sustain dynamically changed trust relationships to adapt various contexts or situations in order to continuously provide system trustworthiness (Yan & Prehofer, 2011). Yan and MacLaverty (2006) proposed that autonomic trust management includes the following four aspects and these four aspects are processed in an automatic way:
- •
Trust establishment: The process for establishing a trust relationship between a trustor and a trustee;
- •
Trust monitoring: The trustor or its delegate monitors the performance or behaviour of the trustee. The monitoring process aims to collect useful evidence for trust assessment of the trustee;
- •
Trust assessment: The process for evaluating the trustworthiness of the trustee by the trustor or its delegate. The trustor assesses the current trust relationship and decides if this relationship is changed;
- •
Trust control and re-establishment: If the trust relationship will be broken/changed or is broken/changed, the trustor will take corresponding measures to control or re-establish the trust relationship.
As we can see from the above, autonomic trust management can be achieved through trust modeling and evaluation.
Various trust modeling and management mechanisms are described in the literature for different systems, such as P2P systems (Kamvar, Scholsser & Garcia-Molina, 2003; Lee, Sherwood & Bhattacharjee, 2003; Liang & Shi, 2005; Singh & Liu, 2003; Song, Hwang, Zhou & Kwok, 2005; Walsh & Sirer, 2005; Xiong & Liu, 2004), e-commerce (Guha, Kumar, Raghavan & Tomkins, 2004), and web services (Resnick & Zeckhauser, 2002). It is widely accepted that trust is influenced by reputation (i.e., public evidence on the trustee), recommendations (i.e., a group of entities’ evidence on the trustee), the trustor’s past experience, and context. Most work focuses on trust evaluation (i.e., trust assessment), but does not consider how to ensure or sustain trust for the fulfillment of an intended purpose. It still lacks comprehensive discussions with regard to how to automatically take an essential action according to a trust value. Although a number of trust models consider the dynamic nature of trust and context’s influence, current literature does not adequately address context-aware adaptation of trust in some domains, such as services, component software, mobile applications, cloud computing and so on (Hall, Heimbigner, Van Der Hoek & Wolf, 1997; Herrmann, 2001; Herrmann, 2003; Malek, Esfahani, Menasce, Sousa & Gomaa, 2009; Mikic-Rakic, Malek & Medvidovic, 2008; Suryanarayana, Diallo, Erenkrantz & Taylor, 2006; Theodorakopoulos & Baras, 2006; Xiong & Liu, 2004; Zhou, Jiao & Mei, 2005; Zhou, Mei & Zhang, 2005). Most of existing solutions focus on a specific system that could be very different with each other. Particularly, due to the complexity and difference of context in different systems, a trust management solution, especially an autonomic trust management solution for one system could become inappropriate for another system. Recently, many solutions were developed for supporting trusted communications and collaborations among computing nodes in a distributed system, e.g., a P2P system (Zhang, Wang & Wang, 2005), an ad hoc network (Sun, Yu, Han & Liu, 2006; Theodorakopoulos & Baras, 2006) and a GRID computing system (Lin, Varadharajan, Wang & Pruthi, 2004). We found, however, that these methods are not feasible for supporting autonomic trust management, since most of them only concern trust assessment and evaluation. Current solutions generally ignore trust control and re-establishment by adapting dynamically changed environments and trust relationships, thus they cannot support or provide autonomic trust management.
The dynamic characteristic of trust is pushing trust management to become autonomic. This requires that trust management should handle the context’s influence adaptively and intelligently. In addition, the trust model itself should be adaptively adjusted to reflect the real situation of a system. Context-aware trust management is a developing research topic and adaptive trust model optimization is an emerging research opportunity.