A new form of computation is emerging rapidly with cloud computing, mobile computing, wearable computing and the Internet-of-Things. All can be characterized as a class of “Cooperative Distributed Systems” (CDS) in open environment. A major driver of the growth is the exponential adoption by people and organizations within all aspects of their day-to-day matters. In this context, users requirements for privacy protection are becoming essential and complex beyond the traditional approaches. This requires a formal treatment of “privacy” as a fundamental computation concept in CDS paradigm.
The objective is to develop a comprehensive formal model for “privacy” as base to build a CDS based framework and platform in which various applications allow users to enjoy the comprehensive services in open environments while protecting their privacy seamlessly. To this end, this thesis presents a novel way of understudying, modeling and analyzing privacy concerns in CDS. A formal foundations and model of privacy is developed within the context of information management. This served as a base for developing a privacy protection management framework for CDS. It includes a privacy-aware agent model for CDS platform with the ability to support interaction-based privacy protection.
The feasibility of the proposed models has been demonstrated by developing an agent-based CDS platform using JIAC framework and a privacy-based Contract Net Protocol. It also included the application scenarios for the framework for privacy protection is Internet-ofThings, cloud-based resource scheduling and personal assistance within the project of smart space.
BACKGROUND AND LITERATURE REVIEW
Despite the comfort that is experienced with new information technologies, they have imposed privacy concerns on people and businesses. The more people engage with digital developments, the more are concerns for their privacy. Primarily, privacy concerns were studied and practiced in legal communities and researches. However, privacy has become inseparable challenge of nowadays digital interactions in which it carries tremendous amount of information about people. Many disciplines have addressed privacy in their solutions however, adequate privacy models for CDS environments is still a challenge.
PRIVACY CONCERNS IN CDS: CONCEPTS AND MODEL
In this work, we concentrate on entities of CDS in an Agent-Based model in which entities can be modeled as CIR agents. These Agents are organized by knowledge, problem solving, interaction, and communication capabilities. “Knowledge” is the entity’s mental state about the world, which is incomplete in many examples of CDS environment and the global knowledge is distributed among all entities. “Problem solving” refers to the entity’s ability to identify the class of their goals, categorizing sub goals, applying required actions to the goals state, and determining the type of interdependency.
PRIVACY PROTECTION MANAGEMENT FRAMEWORK
Figure 3 depicts the operational view of the privacy protection management framework. By applying the risk evaluation model, it is possible to identify the sensitive information that might be shared among entities of the environment while interacting. The messages and sequences of messages among entities construct the interaction protocol of that environment. Providing the framework with the exposure boundary, the interaction protocol, PPL evaluation and the type of privacy protection mechanisms the framework can provide messages and sequences that represent the privacy based interaction protocol.
PRIVACY-AWARE AGENT MODEL AND IMPLEMENTATION
Figure 6 shows the logical architecture of the privacy aware computation entity. Within this architecture, the Exposure Boundary layer collects the exposure boundaries of the information that is shared in interaction protocols.
If privacy solution is not applied the system can reach to a feasible solution. However, by applying privacy protection mechanism, the solution can be reached. For instance the Figure 8 shows a solution that can be approved using privacy protection mechanism applied on information before it goes outside of the exposure boundary.
The proposed privacy protection framework is a generic model that can be used as an analytical tool for identifying privacy states of interaction protocol as well as getting applied in contexts such as computation level. Capturing privacy as a computation concept necessitates incorporating privacy in the computation entity at interaction level. The computation entity in CDS environment requires resolving interdependency problem through interaction.
A PRIVACY-BASED INTERACTION FOR CNP PROTOCOL
The biding structure in CNP includes the identity and node abstraction which include the specification of the node that is providing the proposal. Node abstraction also includes the information that the contractor might need in case of being a winner. In traditional CNP, not differentiating potential contractors and sharing the task announcement discloses sensitive information (Figure 18).
Because the contractor can convey the history of the tasks that are allocated to it and perform operations on it, implicit sensitive information might be transformed to explicit using auxiliary information. This information is shared with the contractor entity in a period of time. To protect this information punishing mechanisms can be more effective. These mechanisms require having an agreement between the participants of the interaction. This introduced the agreement process between the contractor and the requester (Figure 19).
PRIVACY AWARE CDS MODEL: APPLICATION SCENARIOS
Smart space includes entities with various types of capabilities. They are modeled as agents within the environment. Diverse set of devices, sensors and equipment are used in smart space such as kinects, twines, mindstorm, IP cameras, NFC and RFID tags and android-based mobile devices. The logical architecture of the smart space is shown in Figure 21. Many applications and services are created by utilization of these “things” that are registered within this environment.
The approach in resolving the privacy problem as a quality factor in scheduling is intended to limit the solution space for entities that can provide the necessary privacy protection. As shown in Figure 25, the scheduling solution considers all entities as part of the solution space. Then it identifies the entity that can resolve a scheduling request.
CONCLUSION AND FUTURE WORK
The goal of this research is to provide a formal treatment of “privacy” as a fundamental computation concept in CDS paradigm to build a privacy-aware CDS framework and platform. The formal model of privacy served as a base for developing a privacy protection management framework for CDS. It includes a privacy-aware agent model for CDS platform with the ability to support interaction-based privacy protection. Additionally, the feasibility of the proposed models has been demonstrated by developing an agent-based CDS platform using JIAC framework in an IoT-based project of smart space and a privacy-based Contract Net Protocol.
Source: The University of Western Ontario
Authors: Afshan Samani | Hamada H. Ghenniwa