Towards effIcient bloCkchain-based disTributed SDN ArchiteCtures – TIC-TAC-SDN
Towards effIcient bloCkchain-based disTributed SDN ArchiteCtures
There is a growing interest in using blockchain within distributed Software Defined Networking (SDN) architectures to address certain limitations such as state synchronization and trust between distributed SDN controllers. The blockchain technology introduces a significant latency that is incompatible with many Internet of Things applications. TIC-TAC-SDN aims at developing a new adaptive consensus protocol to address this problem.
According to a recent Gartner’s report, one million new Internet of Things (IoT) devices will be sold every hour by 2021, which inevitably comes along with new challenges and demands in terms of high bandwidth, security, ubiquitous accessibility and dynamic management of communication networks. Software-defined networking (SDN) has emerged as a response to the limitations and complexities of traditional network architectures, aiming at consolidating the control over network devices into a logically centralized (software) controller. Nonetheless, the increasing number of smart connected devices in IoT environments implies relying on more than one controller, which is also referred to as distributed SDN (dSDN) in the scientific literature. The use of multiple SDN controllers poses a number of challenges as, for example, the need to reach consensus among multiple controllers, or to deal with security issues including trust and accountability. <br /><br />To overcome these issues, the blockchain technology is receiving a growing attention in the scientific community, as it offers powerful tamper-proof logging and auditing capabilities where trust and control are not anymore centralized and black-boxed, but rather decentralized and transparent (i.e., no need for a central trusted authority). Despite the advantages of blockchain, this technology comes along with constraints, particularly regarding communication delays that may impact on the overall system/application performance and Service Level Agreement (SLA) guarantees. Those delays mainly result from the consensus protocol underlying blockchain, which is used to maintain synchronized all the blockchain copies across the nodes involved in the consensus decision-making process. Given this problem, it is all the more important to come up with innovative and smart ways of evaluating and mitigating those impacts. <br /><br />This is where the TIC-TAC-SDN project comes into play. First, it introduces a rigorous methodology to identify and formalize the impacts of blockchain on the end-to-end dSDN architecture performance in terms of trade-off between convergence time, security and scalability. Secondly, it progresses the current state-of-the-art by investigating, designing and evaluating a novel adaptive blockchain consensus protocol that best meets the trade-off problem. The novelty and originality of the proposal lies in the fact that the consensus protocol is able to self-adapt, at any time, to evolving application needs while maintaining the best trade-off point. In TIC-TAC-SDN, we claim that proposing such new adaptive and smart consensus protocols are key to the success of integrating blockchain with existing systems and network solutions.
To achieve the project objectives that have been introduced in Section I.2, two scientific Work Packages (WP1 & WP2) have been defined, as summarized below:
WP1 – Formal definition of the trade-off problem
Before investigating and proposing any new consensus protocol, it is needed to methodologically and thoroughly study and understand how a given blockchain technology (or family of technologies) impacts on the end-to-end dSDN application performance in order to accurately formalize the trade-off optimization problem addressed in TIC-TAC-SDN. To achieve this, two tasks are defined in WP1, namely: (i) Task 1.1 that aims to develop appropriate Design of Experiments (DoE) in order to identify influencing parameters and interactions considering the architecture as a whole (i.e., considering different types of blockchain technologies/consensus protocols, network topologies, App settings, etc.); and (ii) Task 1.2 that aims to accurately formalize the fitness (optimization) function underpinning the trade-off challenge. To put it another way, a set of blockchain technologies are considered as “black-box” models in T1.1, while T1.2 focuses on studying – for one or a subset of blockchain technologies – the underlying consensus protocol in “white-box” fashion in order to accurately formalize it, which will further help us to formalize the trade-off optimization function. These two tasks are further detailed below.
WP2 – Formal design of a new adaptive consensus protocol:
At this stage, the trade-off optimization problem is formalized, research assumptions RA1 and RA2 verified, and the blockchain-based dSDN platform is available for the coming experiment in WP2. The next step consists now to investigate, formalize, design, implement and verify innovative and smart solutions that enable the dSDN architecture to self-adapt depending on the network conditions and application-specific needs. This corresponds to the “intelligence” that is going to be embedded into the cybernetics’ reconfiguration component. To achieve this, two tasks are defined, namely Task 2.1 that aims to formally design, implement, verify and evaluate a new type of consensus protocol that has the ability to self-adapt, in a real-time and continuous fashion, in order to best meet the fitness optimization function formalized in T1.2; and (ii) Task 2.2 that aims to improve the initial version of the proposed consensus protocol with predictive analytics capabilities. In T2.2, TIC-TAC-SDN aims to explore and develop algorithms to continuously monitor/measure the system convergence time (via South-bound APIs), while taking into consideration network manager needs/expectations (via North-bound APIs) in the view to on-line optimize the fitness function. The goal is to move the towards the co-called Intelligence Defined Network (IDN) paradigm, which aims to apply ML to automatically learn and improve from experience to anticipate network-behavioural patterns/needs.
The first scientific contribution of project is related to the work done by the PI with the recruited MSc student, Matthieu Renard. It gave rise to the following outcome so far
- 1 journal article submitted to Expert Systems with Applications (IF: 8.66), which is now at the 3rd round of review (waiting for result). The submitted article, along with the feedback from reviewers are attached at the end of this document;
- 1 decision-making tool made publicly available online has been released at the following URL: credodlt.sylvainkubler.fr. This tool, which is called CREDO-DLT standing for «multiCRiteria-basEd ranking Of Distributed Ledger Technology platforms« is developed to help decision-maker to select the most suitable DLT platform alternative (i.e., that best suits their needs and requirements). A use case scenario in the context of energy communities is developed in the article to show the practicality of CREDO-DLT (cf., manuscript attached to the report).
After this first contribution, the goal is to start specifying the self-adaptive consensus protocol with the PhD student
- 1 journal article submitted to Expert Systems with Applications (IF: 8.66), which is now at the 3rd round of review (waiting for result). The submitted article, along with the feedback from reviewers are attached at the end of this document;
Software Defined Networking (SDN) and Blockchain are two emerging paradigms in the Internet of Things (IoT). The former greatly simplifies network programmability, the latter provides powerful tamper-proof logging and auditing capabilities that help to solve problems facing distributed SDN (dSDN) such as trust and state consistency issues among SDN controllers. Today’s literature has shown that blockchain technologies can introduce significant latency (>100sec), which is clearly inappropriate for time-critical applications. This may become all the more critical in IoT ecosystems in which hundreds to thousands of SDN controllers/apps need to cooperate/interact with one another. TIC-TAC-SDN is committed to study, design and validate a new blockchain consensus protocol that has the ability to self-adapt to find the best trade-off between reducing the system time convergence, without compromising its scalability and security, and without violating the application Service Level Agreements.
Project coordination
Sylvain Kubler (Centre de recherche en automatique de Nancy (CRAN))
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
Partnership
UMR 7039 Centre de recherche en automatique de Nancy (CRAN)
Help of the ANR 167,011 euros
Beginning and duration of the scientific project:
February 2021
- 42 Months
