Designing for Multiple Blockchains in Industry Ecosystems
Download as PPTX, PDF0 likes124 views
Our proposed method employs a Design Structure Matrix (DSM) and Domain Mapping Matrix (DMM) to derive candidate shared ledger combinations, offering insights into when centralized web services or point-to-point messages may be more suitable than shared ledgers. We also share our experiences developing a prototype for an agricultural traceability platform and present a genetic-algorithm-based DSM and DMM clustering technique.
Designing for Multiple Blockchains in Industry Ecosystems
- 1. Australia’s National Science Agency Designing for Multiple Blockchains in Industry Ecosystems Dilum Bandara, PhD Principal Research Scientist Architecture and Analytics Platform Team CSIRO’s Data61 Dilum.Bandara@csiro.au Cite: H.M.N. Dilum Bandara, Mark Staples, and Sidra Malik. 2025. Designing for Shared Ledgers in Industry Ecosystems. Distrib. Ledger Technol. March 2025. https://doi.org/10.1145/3724410
- 2. Design Science Research Contribution 2 |
- 3. Blockchains
- 4. Goal – Decentralised Trustless Environment Organization 1 Organization 2 Organization 1 Organization 2 Centralised Trusted Authority Traditional trusted environment Blockchain trustless environment Blockchain network Censor & manipulate Single-point of failure Lacks transparency High cost Don’t trust a single node Trust the network: replicated, transparent, & cross- checked data 4 |
- 5. • Main functionally: • Shared append-only database (a ledger) • Shared compute platform (“smart contracts”) • Logically centralises data; administratively decentralises control • Blockchains are good for … • New trustworthy & efficient ways to work together • Exclusive control of digital assets Blockchains Centralised trust using a 3rd party Decentralised trust using a blockchain 5 |
- 6. • Blockchains are one kind of distributed ledger • Ledger is an append-only list of blocks of transactions (TXs) • Ledger structure is one list, but operators are in a peer-to- peer (P2P) network • Distributed ledger systems share ledgers • Can be lots of small ledgers • Perhaps only parties of interest see TXs • Many possible kinds of ledger structures Blockchains vs. Distributed Ledger Technology (DLT) … … … … … 6 |
- 7. • Blockchains are good for multi-party business processes • Maintaining business confidentiality when data on a blockchain? 1. Data segregation 2. Computations on encrypted data Blockchains for Supply Chains 7 | Transparency Smart contracts Nonrepudiation Immutability Accountability Integrity Authenticity Reduced carbon footprint Scalable Availability Traceability Trust Single source of truth Independent verifiability Simplify reconciliation Automation New market models Prevent censoring & cherry picking Data standards
- 8. Ending Plastic Waste Symposium May 2023 • One step up & down focus at every step • Shared access is undesirable • Suppliers shouldn’t see outputs • Customers shouldn’t see inputs • Segregate inputs & outputs • E.g., pairwise blockchain ledgers • Needs a supply chain integrator to get a holistic view • Integrator calculates recycle % & checks mass balances Data Segregation 8 | Suppliers Customer Manufacturer Shared ledger Pairwise ledgers
- 10. Data-Access Matrix of Agricultural Processing Ecosystem 10 |
- 11. • A shared blockchain – One ledger to rule them all • Each other’s data are visible • Even private-permissioned blockchains reveal transacting parties & patterns • Pairwise ledgers • Low integrity as only 2 parties • 𝑂(𝑛2 ) ledgers are economically or administratively infeasible to manage • What are other design choices between these 2 extremes? • Design problem • Design to enhance data sharing in industry ecosystems by optimising the sharing of ledgers using a method that determines what parties should share which ledgers, and what data should be on those ledgers, while satisfying commercial sensitivity needs in the ecosystem Problem Statement 11 |
- 13. Design Structure Matrix (DSM) 13 |
- 14. Domain Mapping Matrix (DMM) • DSM – Party-party relationships • What parties should share which ledgers? • DMM – Party-data relationships • What data should be on those ledgers? 14 |
- 15. Method for Deriving Candidate Designs 15 |
- 16. Design Tactics for Realisation Results from DSM/DMM Clustering 16 |
- 17. • Minimal Description Length Principle (MDL) • Provides a means to encode & quantify degree to which clustered DSM/DMM aligns with original & any residual mismatches • We adopted Yu et al.’s encoding of clusters Clustering Algorithm 17 | Tian-Li Yu, Ali A. Yassine, and David E. Goldberg. 2007. An information theoretic method for developing modular architectures using genetic algorithms. Res. Eng. Design 18 (2007), 91–109
- 18. • Clustering is NP-hard • Used a genetic algorithm • Chromosomes encode cluster membership • Calculated MDL for each membership while considering DLT-specific relationships like source & sink • Keep iterating till you get the lowest MDL value • Code at https://github.com/dilumb/DLTClust Clustering Algorithm (Cont.) 18 |
- 21. Clustering of Agriculture Supply Chain 21 |
- 22. • While integrity of blockchains arises from transparency, it can undercut commercial sensitivity needs of industry • A method to help designers find a middle-way by calculating candidate designs of collections of shared ledgers based on data access requirements • Adapts & extends well-known techniques of DSMs, DMMs, & genetic algorithms to solve a new problem • Candidate designs in seconds to minutes, saving hours to days • Future work • Dynamic ecosystem membership & data access changes • Adopt numerical DSMs & DMMs to represent data volume, or privacy or commercial risk level for better trade-offs & negotiations Summary 22 |
Editor's Notes
- #2: Visual abstract of our design science contributions Design science is a paradigm in applied research, including software engineering It involves designing, investigating, and communicating artifacts within a context Aim is to generate prescriptive knowledge for professionals and share empirical insights gained from applying it in context Design Science Research Methodology
- #4: BC’s goal is to replace the central authority with a decentralised solution based on a set of computers. While central trusted parties such as banks, regulatory bodies, and government are essential to establish trust between 2 transacting parties, they are many concerns around them. For example, They have the power to control and manipulate the system. Lead to a single point of failure. Their internal system state is opaque to the participants. They are fragmented making it difficult to interoperate and collaborate. Leading to high costs. When it comes to the BC network, we don't trust any single computer but trust their collective behaviour. We ensure correct collective behaviour by: Replicating the state across many nodes in the network. Ensure all nodes agree on what changes are made to the state, and by whom. Cross-checking each other’s computations. Making the state public. Also, incentives may be given to nodes to ensure the majority behave correctly. While this appears easy, let's discuss some requirements for replacing the authority and the challenges we need to overcome in doing so.
- #5: Traditionally, when 2 organisations are in business together, often they need to agree to rely on some trusted 3rd party to facilitate the operation of their business relationship. In a blockchain, instead of having to trust some 3rd party, we can choose to trust the decentralised technical infrastructure that provides a shared ledger. To do so, blockchains offer 2 main functions: They are a shared database or ledger to record TXs. They are a shared computation platform for executing smart contracts (SCs). We are already very familiar with databases like relational databases and compute platforms like the cloud. But blockchain is different because it is a distributed system and doesn’t have a central owner or operator. Even though it is distributed, the technical infrastructure ensures that there is consensus among all parties about the ledger and the computation. So, we can say that blockchains logically centralise data but administratively decentralises control.
- #6: When the ledger that stores data is distributed, we call it a distributed ledger. However, compared to multiple databases in a client-server system where data is written only to the master (ready from master and replicas), these ledger copies can be read/written at any node. The Set of technologies around this design is called distributed ledger technology (DLT). So blockchain is one such realisation of a distributed ledger. A distributed ledger system/ecosystem shares multiple ledgers. E.g., 2 parties engaged in a supply chain may run a pairwise ledger and the supply chain may consist of multiple such ledgers. In the bottom-right figure, small horizontal rectangles illustrate (small) ledgers shared between a few parties of interest. As we see in the next session, blockchains can be built in various ways while retaining the idea of a chain/list of blocks. Therefore, a distributed ledger ecosystem may even include multiple kinds of ledgers.
- #7: When we talk about provenance, chain of custody, and traceability blockchain is a technology to consider Blockchains are good for new trustworthy & efficient ways to work together such as multi-party business processes. Supply chains are essentially multi-party business processes Also, unless it is a vertically integrated supply chain, there is no single party to own/control supply chain Hence, blockchain is a good fit for enhancing transparency and accountability in supply chains Blockchains poses a bunch of properties like transparency, immutability, and nonrepudiation that can address many trust, transparency, and traceability issues Properties in dark blue are not perfect but have substantially improved recently Smart contracts enable on-chain accounting (e.g., product carbon footprint (PCF) calculation), ESG ratings, verifiable credentials, auditing, and some aspects of governance They enable independent verification and reduce or remove the need for reconciliation Prevent data censoring & cherry picking Enforce data scandalisation Blockchains no longer have a high carbon footprint are more scalable While these properties lead to many advantages for supply chain traceability systems, a high level of transparency of data stored on a blockchain is in conflict with business confidentiality The rest of the talk focuses on 2 ways of retaining business confidentiality in such a setting These approaches also work on conventional technologies that use some sort of a shared database As 1st low of most technology use, don’t use a blockchain if the problem can be solved with well-established (centralised) technologies. Hence, whether to use a blockchain or not depends on the use case
- #8: When we capture supply chain data we can focus on keeping track of one-step and one-step down interactions If we do it at every point of interaction, we can get a full view of the supply chain However, storing all supply chain data in a single place that everyone can access is not a good idea Suppliers shouldn’t see customer data and vice versa, as it can have implications on price setting Hence, rather than storing both inputs and outputs in the same blockchain ledger, we need to segregate inputs from outputs This retains other properties of blockchain like transparency, immutability, and independent validation between 2 parties having access to the ledger However, we lose a holistic view of the supply chain. Hence, we need a supply chain integrator to connect with each pairwise ledger to get a holistic view of the supply chain For example, this works well in a vertically integrated supply chain That integrator then can calculate recycle % and check mass balances
- #10: Visual abstract of our design science contributions Design science is a paradigm in applied research, including software engineering It involves designing, investigating, and communicating artifacts within a context Aim is to generate prescriptive knowledge for professionals and share empirical insights gained from applying it in context Design Science Research Methodology
- #11: When the ledger that stores data is distributed, we call it a distributed ledger. However, compared to multiple databases in a client-server system where data is written only to the master (ready from master and replicas), these ledger copies can be read/written at any node. The Set of technologies around this design is called distributed ledger technology (DLT). So blockchain is one such realisation of a distributed ledger. A distributed ledger system/ecosystem shares multiple ledgers. E.g., 2 parties engaged in a supply chain may run a pairwise ledger and the supply chain may consist of multiple such ledgers. In the bottom-right figure, small horizontal rectangles illustrate (small) ledgers shared between a few parties of interest. As we see in the next session, blockchains can be built in various ways while retaining the idea of a chain/list of blocks. Therefore, a distributed ledger ecosystem may even include multiple kinds of ledgers.
- #13: Visual abstract of our design science contributions Design science is a paradigm in applied research, including software engineering It involves designing, investigating, and communicating artifacts within a context Aim is to generate prescriptive knowledge for professionals and share empirical insights gained from applying it in context Design Science Research Methodology
- #14: Visual abstract of our design science contributions Design science is a paradigm in applied research, including software engineering It involves designing, investigating, and communicating artifacts within a context Aim is to generate prescriptive knowledge for professionals and share empirical insights gained from applying it in context Design Science Research Methodology
- #15: Visual abstract of our design science contributions Design science is a paradigm in applied research, including software engineering It involves designing, investigating, and communicating artifacts within a context Aim is to generate prescriptive knowledge for professionals and share empirical insights gained from applying it in context Design Science Research Methodology
- #16: Visual abstract of our design science contributions Design science is a paradigm in applied research, including software engineering It involves designing, investigating, and communicating artifacts within a context Aim is to generate prescriptive knowledge for professionals and share empirical insights gained from applying it in context Design Science Research Methodology
- #20: Visual abstract of our design science contributions Design science is a paradigm in applied research, including software engineering It involves designing, investigating, and communicating artifacts within a context Aim is to generate prescriptive knowledge for professionals and share empirical insights gained from applying it in context Design Science Research Methodology
- #21: Visual abstract of our design science contributions Design science is a paradigm in applied research, including software engineering It involves designing, investigating, and communicating artifacts within a context Aim is to generate prescriptive knowledge for professionals and share empirical insights gained from applying it in context Design Science Research Methodology
- #22: When the ledger that stores data is distributed, we call it a distributed ledger. However, compared to multiple databases in a client-server system where data is written only to the master (ready from master and replicas), these ledger copies can be read/written at any node. The Set of technologies around this design is called distributed ledger technology (DLT). So blockchain is one such realisation of a distributed ledger. A distributed ledger system/ecosystem shares multiple ledgers. E.g., 2 parties engaged in a supply chain may run a pairwise ledger and the supply chain may consist of multiple such ledgers. In the bottom-right figure, small horizontal rectangles illustrate (small) ledgers shared between a few parties of interest. As we see in the next session, blockchains can be built in various ways while retaining the idea of a chain/list of blocks. Therefore, a distributed ledger ecosystem may even include multiple kinds of ledgers.