- LE Kim hung
Thesis
Interoperability Mechanisms in Industrial IoT applications and Smart City
Industrial Internet of Things (touted as Industry 4.0) is now limited to a set of vertical silos benefitting little from surrounding connected systems or applications to enrich the measurements, analysis and thus the quality of decisions. In this context, the state of the art is closer to the M2M system promises of the Internet of Things. For example, a system for measuring the air quality on an industrial site will not benefit from weather data or data from a capillary network as PACA AIR to enhance the relevance of its measures. Similarly, a remote control M2M network of industrial facilities not today constable to CRM or ERP standard tools to fit into the management of operational actions of the company.
We identified that the locks of interoperability lie in the radio access network, data collection protocols, devices management protocols and APIs to access databases or "data lake".
The purpose of this thesis is to rely on the new standards such as LwM2M, MQTT to demonstrate the feasibility and gain interoperability in Industrial IoT applications and smart city.
The expected outcome of this work will be incorporated into an industrial perspective. The thesis will explore a number of problems and their solutions. The proposed solutions will be pushed into prototyping stages and advanced experiments that will take into account both the research and industrial constraints and get ground returns.
The non-exhaustive thematic exploration will include:
· Exploring a convergence layer of LPWAN standards (LoRa, SigFox ...) and cellular (GPRS, LTE-m) on an agnostic device management layer on the radio access and data collection protocols (MQTT, CoAP).
o Research problems - How will upcoming technologies like LoRa, SigFox will interoperate with cellular technologies? It is critical to investigate which use cases of IoT requires such convergence layer and also the requirements of the layer. Interoperation of the protocols like MQTT and CoAP also needs to be studied.
§ For uniform device management, a common set of vocabularies are necessary which is not available (at the moment) in any Standard Development Organizations (SDOs) [1].
· Exploring "backend" architecture for verticalization of generic use cases. Indicators customization methodology for building dashboards "trades". Setting indicators manufacturers models consist of data vectors creating modules from aggregated data sources and metrics modules to be applied.
o Research problems
§ The main challenge here is how to create an interoperable and generic architecture that will conform with ongoing standardization efforts [2].
§ Uniform treatment of data coming from heterogeneous sources which will lead into actionable intelligence [3]. The current industrial and academic approaches do not converge on a unified data analysis. This thesis will explore two possible solutions - (i) use of semantic web technologies to settle the heterogeneity at data processing level and (ii) investigate the use of machine learning for IoT and advanced data analysis.
§ The results of above results should be integrated into a prototype for a generic use case.
§ Scalability of the architecture will be thoroughly examined when deployed to an M2M gateway and/or a remote cloud system. The research opportunities are to identify which contexts need a gateway based deployment and which one requires cloud. This has the potential of developing a "best-practice" or "guidelines" for industry related to the scalability of the architecture.
§ In addition to that, edge computing could also be researched. This paradigm allows the cloud computing facilities to be pushed towards the edge of the cellular networks which provides better QoS, reduced delay and improved UX.
· Interoperable architecture exploration aggregation "time stamped" Geo data and localized "semantics". The data can come from server applications, M2M device or mobile. The deployed sensor networks must be able to enrich the data with existing networks but also by user mobility.
o Research Challenges
§ Interoperability needs to be considered at several level - (i) data collection, (ii) data storage and retrieval, (iii) data dissemination.
§ Sensor Markup Language (SenML) can be used to get timestamped data coming from heterogeneous sources. SenML extensions provide a unique way to interact with actuators [4], [5]. But the challenge is to create a lightweight software that can produce such metadata from Class 0 type devices. JSON based implementation might not be sufficient for this. The thesis will look into alternate implementations like EXI.
· It is equally important to include a module for discovery in the overall IoT ecosystem to allow high level applications as well as consumers to look for the required M2M device or services. To realize the global vision of Web/Internet of Things, there must be mechanisms to discover resources, their capabilities and properties. Thus resource discovery becomes a fundamental requirement of any IoT platform. A gap analysis of current research directions and standardization efforts reveal: (i) there is no common catalogue to describe the resources, units and domains, (ii) synonyms are also often not recognized. Therefore, searching for a resource using a synonym might not discover the resource, (iii) the content of discovery result metadata, ranking of the discovered resources are not standardized, (iv) there is a lack of uniform mechanism for resource discovery. As a result, interoperability among WoT/IoT platforms are not maintained, (v) there are no uniform guidelines about how to integrate adequate security mechanisms and access control functions into a discovery framework, (vi) the semantic based discovery is not studied in depth yet [6].
o Research Challenges
§ To create an API which supports uniform and technology independent discovery of smart things and services.
§ The thesis will explore the following in relation to this work item - (i) Provide guidelines for binding to APIs and protocols for efficient interaction with things, (ii) Utilize a uniform catalog of descriptions (as defined in things description) for discovery of things and their metadata, (iii) Provide means of ranking the discovery results (when presented to end users or other M2M applications).
§ How to securely manage the authorization of thing discovery.
· Security, Privacy and Trust - Client-side APIs that supports the integration/utilization of security and security-enabling functionality by utilizing a common client-side stack [7].
o Challenges
§ IoT application developers have to either create secure code by themselves or re-use provider-proprietary libraries. The latter is further troubled by the fact that there are not yet widely accepted standards for doing authentication and authorization in IoT => this practice leads to silos. The challenge is to create a SPT framework which will be open source and interoperable.
· The process of standardization [2] of the Internet of Things is being formed, it will take several years to complete. This thesis will be the opportunity to see the limits of the proposed standards for the particular context of industrial applications and be proactive.
