Edge Analytics (Deep Learning on the Edge) with Blockchain

#DXC Technology #Blockchain #CBA #AI # Deep Learning

Edge Analytics, AI, ML, Deep Learning, Blockchain - they have become buzz words of every conversation you try to have nowadays. NAB slashing 6000 jobs, AI Chatbots taking over Service Desk Agents jobs, etc. all of this can't be bad for society, can it? 

Let's start with a use case of Edge Analytics and Blockchain and see where we land:

Use Case: Leakage in Gas and Utilities

Participants: Utilities Provider + Insurance Company + Maintanence Company.

Utilities providers supply 24x7 Gas & Electricity to millions of users and due to the nature of service, they have to ensure that even a small leakage is detected in real time otherwise loss to society is huge. Using Deep Learning at the Edge, any leakage can be detected in real time.

Scenario – Utilities Provider X supplies Gas and Electricity 24x7 to Sydney, Australia. On a random day, a gas pipe starts leaking in a Residential and Commercial suburb of Sydney, but luckily provider X has Edge Computing in place so its Deep Learning on the Edge device detects the leakage right away and sends analytics to all relevant parties through Blockchain. CorDapp on Blockchain sends out a service request to the Maintenance company and maintenance company will then notify one of its field technicians to go in and fix the leakage. In parallel, CorDapp on Blockchain also lodges a claim with the Insurance company on behalf of Utilities provider X for the damage cause by the leakage. As this is a permissioned ledger, Insurer and reinsurers (if applicable) will have access to the data related to the incident and can finalise the claim in real time throuh Smart Contracts, i.e. a simple business logic of "does the incident match with the provisions of claims provided in the Policy". Based on the logics defined by the parties, Smart Contract will reimburse Utilities provider X on behalf of the Insurer and Maintenance company’s invoice will be automatically paid as and when it’s due, thanks to Smart Contracts’ self execution feature.

This is an example of savings of millions of dollars for all parties involved and uninterrupted service for Society by simply receiving information in real time via use of Deep Learning on the Edge and simpligying processes through Blockchain technology.  

Assumptions –:

• Utilities Provider, Maintenance Company, and Insurance Company use Blockchain to trade and have Smart Contracts in place.
• Tokens are converted into Feat currency by using New Payment Platform (NPP) or Crypto Currency is used to transfer value among parties.
• All parties use Corda Blockchain and CorDapp is set up as a Front End tool to transact.
• #Greenwave Systems Inc. is the IoT Platform provider.

Let's deep dive into what is Deep Learning on the Edge

Definition: At the core, "Analytics" is the discipline that applies logic (that is, "rules") and mathematics ("algorithms") to data to provide insights for making better decisions. "Edge" analytics or DeepLearning on the Edge means that the analytics are executed in distributed devices, servers or gateways located away from corporate data centers or cloud servers closer to where the sensor data is being generated.

Why Edge? - The edge is where people and physical things connect and start to converge with the digital world. Personal interactions with things are becoming more ambient (e.g., voice processing or processing based on location), and even immersive (especially augmented and mixed reality). A business moment could be the point at which a person is walking by a display, driving by a business or in a room full of constituents. Those simple acts may be catalysts for unknown opportunities for a personalized product or service. A tremendous amount of intimate data is being produced that may be valuable to a person or within a location.

However, it may not be valuable enough or it might be too intimate to send to a central core — e.g., when someone leaves the house, falls asleep, goes to the restroom or goes to the doctor. At the same time, immersive technologies are beginning to enable new business moments and interactions that need to be real-time and may be fleeting (e.g., when a known frequent shopper walks by a store, or looks at an item on the store shelf). Immersive technologies have enabled more real-time human and human-to-thing interaction, and that will require shared processing power and storage close to the edge.

Position and Adoption Speed Justification: There has been inflated expectations due to increased awareness and understanding of the importance of the technology by both end-user organizations and vendors. Proliferation of IoT devices and the need for real-time insights are the greatest drivers of analytical computing at the edge of the network. Consequently, as IoT becomes mainstream, enterprises will demand more edge computing to improve the real-time analytics and drive business process optimization. More IoT platform and analytics vendors are adding the ability to deploy and run small footprint analytics packages on edge devices. It reflects the shifting balance between edge computing and cloud computing which is one of the most significant trends in IoT. This trend is being driven by use cases where data needs to be analyzed in real-time and the realization that moving data back and forth between remote locations is far too costly a practice. It also is about making the edge devices more intelligent.

Prime Use of Edge Analytics: Analytics Leaders should consider edge analytics when:

• A factory, vehicle, home or other distributed site must operate even when disconnected from a cloud platform or corporate data center.
• Regulations or laws require that data be kept within the country or other locality where the data is generated.
• The network does not have the capacity to upload all the data to a central location in a timely fashion, or when it would cost too much to upload all the data, and when there is no benefit to moving the details of the raw data to the central location.

Business Benefits/Impact: Running analytics at the edge will become the new normal in IoT architectures by the time it reaches the plateau.

Not all data needs to be sent to the cloud or main data center, because it's cost-prohibitive or bandwidth-intensive, it has performance implications or it's impractical in remote locations.Therefore, it's imperative to have some components of the IoT solution deployed at the source of data generation as an aid to rapid decision making using real-time analytics.

Edge computing solutions may vary from basic event filtering to complex-event processing or batch processing. Systems with adequate computing power must be deployed to address these data processing requirements. Figure at the top represents the categories of edge computing based on computing capabilities, and a combination of these may be deployed to create a comprehensive edge computing solution.

Advantages include:

• Faster response times — For example, applications such as leak detection in utilities and oil and gas require response times in seconds. When that data is sent to a central location for analysis, delays are introduced and it loses its value for real-time requirements.
• Reduced network bottlenecks — For example, video data employed in smart city applications, such as traffic management, could be so large that it could congest the network if all the data needs to be transferred to a centralized location for analysis.
• Data filtering — This reduces the data management and storage overhead by using edge analytics to look for just the actionable data. As a result, only the necessary data is analyzed or sent on for further analysis.
• Reliability — the remote location can remain in operation even if the network, cloud servers or data centers are unavailable.

Disadvantages include:

• Increased complexity — Remote, distributed analytics deployment and management make the deployment and management more complicated than for aggregated data in a single location.
• Increased security risk — There is potentially increased cybersecurity vulnerability by locally persisting device data and analytics.
• Reduced data granularity — There is a potential loss of useful insight by "throwing away" raw data stored locally.

Intelligent Edge Use Cases

• Manufacturing Yield Optimization
• Pump Cavitation Detection
• Remote Monitoring for Transportation
• Wind Turbine Optimization
• Pipeline Optimization
• Condition-based Pump Maintenance
• Predictive Maintenance for Power Plants
• Streetlight for IoT Sensors
• Intelligent Manufacturing
• Wind Power forecasting
• Mining Optimization
• Intelligent Real-time Health Monitoring
• Leakage in Gas and Utilities
• Areas where Laws prohibit data transfer

About the author: I work for one of the world's biggest IT Service Provider and I am fortunate to work on Australia's number one Bank. Blockchain, Artificial Intelligence and its sub categories are here to make our life easier and will benefit Society. My suggestion to all of you is summarized beautifully below.

"Afraid not, Embrace the change. It's the resistance to change that's painful not the change itself."

                                                          Budha