METRICS - prograMmablE datacenter foR Iiot appliCationS

Project Website: ---

METRICS - prograMmablE datacenter foR Iiot appliCationS

Start Date:
July of 2016

Industrial Internet of Things (IIoT) is a recent trend in industry to adopt IoT technologies and, in that way, further increase flexibility and dynamism at reduced costs for manufacturing processes, exploiting the dynamism of the ICT world. Such an approach envisages enhancements to typical industrial scenarios by monitoring the behavior of people and equipment (things), allowing the creation of real-time data-driven decision making systems that overcome the need for human intervention. In such an IIoT scenario, the manufacturing equipment will be coupled with sensors and actuators connected through local networks to SCADA (Supervisory Control And Data Acquisition)servers and now, more commonly, to a local data center in a Cloud Computing environment (also known as Fog Computing). The datacenter will provide specialized tools that collect and analyze large amounts of data, determining trends and setting new production strategies enforced by local resource managers that apply the most effective configuration to the production elements.

The development of ICT technologies has been driven, up to a large extent, by the integration of communications, from conventional telecommunications to computer middleware, multimedia systems,etc, to enable storage, access and exploration of information. Conversely, the focus on integration has led to relaxed Quality of Service (QoS) and even poor dependability in some cases, visible in the current Internet, which clashes with the requirements of industrial systems, often containing critical components with strict predictability, dependability, latency and timeliness needs. For instance, common industrial scenarios frequently include closed-loop control systems with cycle times below 1 ms, in which message losses or late message delivery may induce disturbances that can result in important economic losses. Therefore, the development of the IIoT concept implies evolving the existing ICT technologies so that they can guarantee the performance required by critical components of industrial systems, while at the same time maintaining their native advantages, particularly concerning convergence.

Software Defined Networks (SDN) is a relatively new network management architecture hat is receiving growing attention from industry and academia, exhibiting two features that are very well suited to manage a network with real-time requirements: (i) centralized resource control completely decoupled from the data plan; (ii) fine granularity in resource control, down to validating each single packet, received in each port of a switch. Despite offering unprecedented control granularity, the guarantees offered by SDN are still insufficient to enable real-time connectivity. Querying the vast literature about SDN-based QoS, to the best of the research team’s knowledge, there is no publicly known work about real-time SDN, and even its signaling protocol, OpenFlow, lacks mechanisms to specify real-time requirements.In order to achieve the desired communications QoS, significant research and development efforts focus only on the data link layer technologies. In particular, Ethernet has been receiving substantial attention and several enhancements were already proposed to achieve the desired support for real-time traffic.

Several efforts have also been made recently to grant real-time features to wireless communication, particularly using IEEE802.11 (WiFi). Some of these enhancements have been carried out within IT, such as the HaRTES switch, which is a real-time Ethernet technology that provides dynamic configuration of allocated network resources. Our main objective is, thus, to enable the IIoT by leveraging SDN in industrial systems, supported on appropriate data link layer technologies that provide the needed QoS enhancements, thus allowing direct equipment to equipment real-time interaction but also real-time interaction with the internal cloud (data center). The HaRTES switch will be instrumental to our goal, to enforce real-time communications on Ethernet and also on WiFi access networks, according to high level network management carried out by SDN.

Therefore, the specific objectives of the project are:

  • Integrate HaRTES real-time switching platform with OpenFlow protocol;
  • Integrate HaRTES with WiFi Access Points;
  • Propose enhancements to the OpenFlow specification to include real-time features;
  • Carry out real-time management of network resources with real-time enhanced SDN;
  • Showcase SDN real-time capabilities, disseminating OpenFlow within the industry, particularly with the project partners Altran and Altice;

The project will developed by members of three IT groups, belonging to two IT sites: Telecommunications and Networking – Av, Embedded Systems – Av, Network Architectures and Protocols – Av and Networked Systems – Po, in collaboration with two external companies (Altice Labs and Altran).

The project has started on July 2016, going all the way to June 2018.
The team:

  • IT Aveiro: Pedro Gonçalves, Paulo Pedreiras, Joaquim Ferreira, Francisco Fontes
  • IT Lisboa: Luis Almeida, Sérgio Crisóstomo, Rui Prior

The Tasks

  • Task 1: Requirements specification, architectural definition and modeling
  • Task 2: HaRTES integration with OpenFlow protocol
  • Task 3: OpenFlow real-time enablement
  • Task 4: HaRTES admission control algorithm port to OpenFlow controller
  • Task 5: Extension to WiFi peripheries
  • Task 6: Real-time load-balanced resource management
  • Task 7: Prototype deployment and result dissemination