Blockchain, Machine Learning and its impact on the Digital Ecosystems
Blockchain is a disruptive technology that is taking the digital world by storm and transforming its trusted framework from a centralized to a decentralized one. Architectures built around a centralized trusted framework expose them to a single point of attack and failure. There is no standard way of doing end-to-end encryption in these architectures and hence they are open to variety of security threats such as malware, man in the middle attacks and application hacks that can expose sensitive and private data.
Blockchain on the other hand provides a decentralized trusted framework. It is basically a distributed public ledger of records of digital assets. The main characteristics of the Blockchain technology are distributed consensus and anonymity. Each record of the digital asset in the public ledger is entered by the distributed consensus of the majority of the participants in the Blockchain system. All digital record entries in the ledger are cryptographically linked to each other in a chain so that digital record once entered can never be erased without changing all other entries. It is both computationally difficult and impossible to achieve in a distributed trusted framework. To use a basic analogy, it is easy to steal a cookie from a cookie jar, kept in a secluded place than stealing the cookie from a cookie jar kept in a market place and being observed by hundreds of people.
Figure 1. Scalable and Fast KSI Blockchain Technology Implementation.
Traditional Blockchain technologies widely used in crypto-currency approach suffer from scalability and settlement time issues. The size of such Blockchains grows linearly with the number of transactions. Also the need for computationally intensive Proof of Work means that a Block of records can only be added in 5-15 minutes. This may be too high to be useful for many real-time applications. However, settlement times can be reduced significantly by eliminating the need for Proof of Work by limiting the number of participants in the Blockchain ecosystem and achieving consensus synchronously. Such an approach is used by Guardtime in its KSI (Keyless Signature Infrastructure) Blockchain Technology.
Machine Learning (ML) describes software that gives the computers an ability to “learn” without being explicitly programmed (Arthur Samuel-1959). ML software is self-adaptive and continues to improve its prediction as it captures new information or training set. A good example of ML is a spam email filter that improves its own ability to spot junk email over time. The accuracy of the prediction by ML software depends on the accuracy and trustworthiness of the training set. Another useful concept regarding ML is Anomaly detection, which finds the unusual, catches the fraud and discovers strange activity in large and complex dataset.
In traditional centralized architecture, ML based batch data analytics takes place in the cloud. To improve performance, IoT networks are adopting fog computing models that extend the cloud capabilities to the edge of the network. Such architectures are prone to attacks at one or more nodes.
The Blockchain based ecosystem makes available “Trusted Copy” of the database to every participating node. The trustworthiness of every record in the database is achieved through a consensus protocol and can’t be modified. Contrast this with centralized architecture, where the aggregated data is available only at specific nodes. It is responsibility of these nodes to maintain the integrity of the data available to them. In the distributed architecture any node can run batch data analytics on its “Trusted Copy” of the database.
Smart Contracts (SC) concept was the invention of Nick Szabo (1994) and is the killer application that runs on the top of the Blockchain. SC are self-executing contactual states stored on the blockchain, which no body controls and therefore everyone can trust. They can be combined with Anomaly detection capability of a ML system to send critical messages. For instance, a healthcare platform can use this to send secure messages to emergency response system if a life threatening health event (e.g. very high blood pressure) is detected. Similarly Industrial IoT systems can send critical alert messages ( e.g. fuel leakage) so that remedial action can be immediately taken.
Emerging Digital Ecosystems
Traditional cloud based ecosystems are increasingly being challenged by distributed solutions based on the Blockchain.
Figure 2. MetaDisk Model Of Data Storage.
For example, existing cloud based storage solutions such as Dropbox, Google Drive or One Drive are being challenged by Storj which provides peer-to-peer cloud storage network that uses MetaDisk, a Blockchain based decentralized file storage application. This solution does not suffer from issues like security, privacy and data control of traditional cloud based storage solutions.
Figure 3. Centralized IoT Networks to Open Access IoT Networks.
IoT is increasingly becoming popular in both consumer and industrial space. A vast majority of IoT platforms are based on a Hub and Spoke model, where a broker or hub controls the interactions between the devices. However, this approach has become impractical for many scenarios in which devices need to autonomously exchange data between them. The Blockchain technology facilitates the implementation of decentralized IoT platforms, which store record of all messages exchanged between smart devices in a decentralized IoT topology. IBM has developed an experimental ADEPT platform that uses the Blockchain technology to build a distributed system of devices. Filament is a startup that provides a decentralized IoT software stack that uses the Blockchain to store unique identities of the devices.
Blockchain technology is revolutionising the digital ecosystem by offering distributed trusted framework. Other emerging technologies such as Smart Contracts and Machine Learning can be layered on top of Blockchain to realize secure, efficient and cost effective digital ecosystems.
Originally published here.