Cloud providers are today enriching their services to support smart Internet of connected objects, or Things (IoT), e.g., the IBM's Watson IoT platform or the Amazon's IoT service. This cloud only solutions already provide valuable supports to connected objects. However, meeting the demand of real-time computation power, that will be generated by the estimated billions of smart objects deployed in the near future, may require to make some structural architectural changes to this cloud-only paradigm.
Fog Computing is considered a key enabler for scenarios where such centralized cloud-based platforms are impractical. Fog computing is a hierarchical networked system paradigm, in which computation and storage resources are placed in proximity to IoT devices.
Fog computing is not intended to replace the current cloud-only architecture, but rather to complement it and improve the computation response time. The fog layer acts as an intermediary layer between the end user and the cloud layer, with the goal of processing user requests locally thus reducing the need of cloud services. Accordingly, a key measure of a well designed Fog architecture is the fraction of computations that can be executed directly in the Fog layer, i.e., without the cloud service, that should be as high as possible.
To achieve this goal, cooperation among fog providers or among fog nodes of the same provider is considered a required feature. Enabling inter-provider collaboration, however, is a multi-facet problem as issues concerning privacy, incentive to cooperate, designing distributed cooperating algorithms have to be addressed globally. The aim of this research project is to focus on this problem of cooperation among competitive fog providers, and has the following two objectives:
Objective 1. Understanding and quantify the benefit of cooperation among fog providers.
Objective 2. Propose and evaluate distributed and secure cooperation algorithms.
FOGCOOP will provide a comprehensive view of the multi- faceted problem for enabling cooperation in multi-provider fog computing environments. Fog computing is considered a strategic step to support the deployment of IoT: Cooperation among providers is a poorly addressed topic in this area, with the potential of allowing elastic computation services and improve the QoS considerably.
Our preliminary study (Objective 1) is expected to provide theoretical evidence of the advantage of cooperation, the risk connected to cooperation, and the principles that have to be used in designing effective cooperation algorithms. Indeed, some results on this direction obtained by members of this proposal has recently be published by a leading journal in this field [see publication list], confirming that it is considered a relevant topic. The expected results of the project will contribute to advance the current state of art in the way cooperation has be addressed in the context of Fog Computing. We intend to employ state-of-the-art game theoretic techniques and techniques inspired by Economics and Markets, such as Mechanism Design and novel models of ¿altruism¿ in societies. We intend to augment current approaches by novel modeling.
Cooperating nodes in the studied Fog computing model may in general belong to different fog providers. As so, they act autonomously and are closed to other competitors. However, much like ¿roaming¿ increased the end-user service level in the cellular telephones and eventually the rewards of different players, cooperation among fog nodes may achieve similar benefits. Today's designs and deployment tend to focus on infrastructure redundancies across tenants; we argue that a collaborative architecture can bring down the costs of IoT deployments and improve efficiency in edge environments. Moreover, computation and data sharing could enable richer, more interesting services and interactions between service providers. The question remains; can this cooperation gain become a major incentive for providers to collaborate and find security, privacy, and payment mechanisms to efficiently implement cooperation? We hopefully will provide a positive asnwer to this question.
FOGCOOP will propose distributed secure incentive based cooperating protocols (Objective 2) designed on the basis of the outcome of Objective 1. We guess that a closed micropayment system among providers can be used as a mechanism to stimulate cooperation in a fair way and with a bounded risk. The algorithms will be inspired to game theory. To the best of our knowledge, in the fog computing research literature this approach has no yet investigated. The achievement of Objective 2, with the support of experimental evidence will hopefully contribute to show the effectiveness of secure cooperation among fog nodes.