Access Control for Distributed Ledgers in the Internet of Things: A Networking Approach

Andrew Cullen, Pietro Ferraro, William Sanders, Luigi Vigneri, Robert ShortenIEEE Internet of Things Journal 2022 · DOI

TL;DR

This paper introduces a novel access control mechanism for Distributed Ledger Technologies (DLTs) specifically designed for the Internet of Things (IoT). By adapting networking concepts like TCP's rate control to a reputation-based DAG ledger, it achieves efficient, fair, and secure transaction throughput management without the energy waste of traditional Proof-of-Work systems.

What Problem Does This Paper Address?

The Internet of Things (IoT) is rapidly expanding, creating a need for decentralized systems that can handle a massive number of devices and transactions securely and efficiently. Traditional DLTs, like blockchains, were not built with IoT constraints in mind. They suffer from low transaction throughput, high latency, and significant energy consumption due to mechanisms like Proof-of-Work (PoW). This makes them unsuitable for resource-constrained IoT devices and large-scale networks. The paper addresses the challenge of creating an access control system for IoT-friendly DLTs (specifically DAG-based ones with reputation systems) that can manage transaction rates fairly and securely, ensuring network resources are utilized effectively without being overwhelmed by malicious actors.

Methods

The core of the paper's contribution lies in its novel access control algorithm, which integrates networking principles into DLTs. The approach has three main components:

  • Scheduling (DRR-based): A modified Deficit Round Robin (DRR) scheduler is used to ensure that honest nodes' transactions are not unduly delayed by congestion caused by malicious nodes. This scheduler prioritizes transactions based on node reputation, ensuring fair access.
  • Rate Setting (AIMD-inspired): Drawing inspiration from the Transmission Control Protocol (TCP), an Additive Increase Multiplicative Decrease (AIMD) algorithm is adapted to allow "best-effort" nodes to dynamically adjust their transaction issue rate. This helps utilize network capacity efficiently while preventing excessive congestion.
  • Buffer Management: A scheme to manage node inbox buffers is implemented to prevent malicious nodes from filling buffers and causing honest transactions to be dropped. This component acts as a safeguard against excessive malicious behavior.

The algorithm is designed for DAG-based DLTs, which offer higher throughput and lower latency than traditional blockchains. It replaces PoW with a reputation system, making it more energy-efficient and suitable for IoT.

Datasets

Not discussed in the paper. The evaluation uses a Python simulator for DAG-based distributed ledgers.

Key Insights and Ideas

  • Networking concepts are applicable to DLTs: The paper demonstrates that established networking principles, like TCP's rate control, can be effectively adapted to solve DLT access control problems, offering a new paradigm.
  • Reputation as a primary access control mechanism: Moving away from energy-intensive PoW, the paper leverages node reputation to fairly allocate network resources, making DLTs more sustainable and accessible for IoT.
  • DAGs are a better fit for IoT DLTs: The inherent structure of Directed Acyclic Graphs (DAGs) allows for higher transaction throughput and lower latency compared to traditional blockchains, making them more suitable for the demands of IoT.
  • Fairness and Security can coexist: The proposed algorithm successfully balances the need for fair resource allocation (based on reputation) with robust security against malicious actors, a critical challenge in decentralized systems.
  • Adaptive rate control is key for efficiency: The AIMD-inspired rate setting allows nodes to dynamically adjust their transaction issuance, maximizing network utilization without causing detrimental congestion.

Key Results

  • High Resource Utilization: The algorithm achieves close to 100% utilization of node resources up to the defined global writing power limit (v).
  • Low Latency: In honest environments, the mean latency converges to a steady state of around 5 seconds, with IoT nodes experiencing lower latency than value nodes.
  • Fairness in Dissemination Rate: Simulations show that nodes receive access to write to the ledger proportionally to their reputation, demonstrating effective Sybil protection.
  • Robustness to Malicious Actors: In a malicious environment, the buffer management component successfully penalizes malicious nodes by dropping their transactions, preventing them from disrupting the network or gaining an unfair advantage.
  • Parameter Robustness: The algorithm demonstrates robustness to changes in parameter choices and can scale with the number of nodes without requiring re-tuning.
  • Comparison to PoW: Simulations show that PoW is highly sensitive to difficulty settings and leads to significant underutilization or latency blow-up, while the proposed algorithm maintains high utilization and stable latency.

Limitations and Future Work

  • Approximate Fairness in Latency: The paper acknowledges that approximate fairness in latency is achieved, with lower reputation nodes sometimes experiencing slightly higher latency.
  • Experimental Results: The paper relies on extensive simulations rather than theoretical proofs for its validation.
  • Future Work: The authors plan to deploy and test the algorithm in the IOTA Foundation's GoShimmer test network and subsequently on the main network. Further research will focus on mathematical analysis of the algorithm and improving scalability to accommodate the growing number of IoT devices. They also plan to investigate the impact of varying network topologies and explore more sophisticated buffer management and scheduling strategies.

Summary updated

BibTeX
@article{cullen2022access,
  author = {Andrew Cullen and Pietro Ferraro and William Sanders and Luigi Vigneri and Robert Shorten},
  title = {{Access Control for Distributed Ledgers in the Internet of Things: A Networking Approach}},
  journal = {IEEE Internet of Things Journal},
  year = {2022},
  doi = {10.1109/jiot.2021.3096129},
}
Figure