SDN: Getting Started

We understand the network level of an IoT system as the protocols, and devices implementing these protocols, that operates on layers 2 and 3 of the OSI stack. At this level, we have to contemplate pervasiveness and ubiquitous network aspects as we are not working with traditional networks. The particularity of IoT networks is the implementation of different types of protocols adapted to the needs of the devices and data types to be communicated. Moreover, the integration of new devices on the architecture, that need to communicate with each other, make the interoperation on the N2N layer more difficult. In this case, we describe `network interoperability` as the quality of communicate devices belonging to different networks or different sub-areas of the same network that belongs to an IoT deployment. Also, the operations on highly constrained environments are also an important issue to analyze. And finally, the use of really heterogeneous protocols (6LowPAN, RPL, LoRa, SIGFox, etc) and mechanisms (tunnelling mechanisms over IP, GRE and 6LoWPAN, etc) at IoT network level are problems we desire to solve with a proposed interoperability solution. The interoperability solution will be based on software defined paradigms but mainly on two approaches: SDR for interoperability on access network and SDN/NFV for the core network.

Common Approaches

To achieve network to network interoperability, the solution we propose is based on the paradigm of Software Defined components, and extension of its capabilities as:

  • Decoupling of data plane from logical plane using a well-studied protocol.
  • Virtualizing network services at the top of the architecture.
  • Implementation of techniques for traffic engineering or QoS to handle different flows of data.
  • Replacing discrete analogy radio components by digital signal processing (DSP) in software.

QoS

QoS is defined as the minimum values of different parameters that an application require to the network for its good performance. These parameters include i.e. low latency, high packet delivery ratio, low packet loss etc. And QoS is the term to represent this capability of networks. QoS allows the coexistence of new applications with different requirements over the same networks. It helps network schedulers prioritize traffic flows according to their requirements. Even though, many technologies i.e. DiffServ or IntServ, emerged to supported guaranteed quality of service for certain applications, e.g. video conferencing, they are not extensively used. Progress in L1&L2 network engineering allowed overprovisioning of resources which made those technologies in many cases obsolete. Moreover, as network diversity was increasing, ensuring end to end latency across different protocol stacks and devices was becoming more and more difficult. Yet, QoS is coming back with higher intensity as Industrial Internet of Things introduces new possibilities and challenges. Despite the business opportunities this may entail, new challenges also emerge. For instance, remote control of life-critical equipment has stringent latency and PDR requirements and is usually implemented over best effort networks like IP. The traffic volume of those applications is low but every packet’s delivery is critical. IoT network then, may be either managed or unmanaged. Managed networks can provide certain QoS or an estimate of it, whereas unmanaged networks lack this predictability. Thus, overprovisioning, being an option is not a desirable feature as it translates to more energy consumption, which the low-power small devices used in critical applications, cannot afford. Therefore, INTER-IoT should focus on gateway and network level to address the issue of QoS.