When is decentralized processing of data the right way to go?
The rapid development of 5G, the impending 6G, and IoT have brought a lot of attention to the decentralized or edge processing of data. Similarly, advancements in autonomous vehicles have increased the need to swiftly and precisely utilize data at the edge to determine the right actions. We can observe the processing of data outside of the central server (in the so-called “user’s edge” or in a decentralized manner) in some personal data use cases. In this article, we look at some of the shared characteristics and when decentralized processing may unlock more use cases.
What Is Edge Processing?
At its core, edge processing refers to processing the data on an edge device, for example, a user’s cell phone, computer, or car, or a network node that is closer to the user than a centralized infrastructure, e.g. cloud. These nodes or user devices also generate data, which combined with other data sets available for the user could be combined and utilized to derive a certain insight and suggested action.
Consider the example of an avid wearable enthusiast who gets alerts when his heart rate or stress levels, measured either by HRV or cortisol level measurements (note: one of our partners has an amazing device for these purposes, reach out to us for more information) are elevated and suggestions on how to regain balance, potentially utilizing historical data for what worked in the past to lower stress for the same individual, a prior known data point.
The phone would access the wearable device via Bluetooth, combine the data readings with the triggers on the phone and send out a push notification or text message to the user. No external server is used and no external log of the user’s behavior is recorded.
A self-driving car functions in a largely similar way, comparing existing data to its lidar sensors’ readings and suggesting the appropriate next actions, in real-time.
Where Does Edge Processing Matter In the Personal Data Space?
There are several technical reasons where edge processing of data is important and there are several business cases. They largely break down into the following categories:
- Continuous data availability is required to generate real-time action;
- Using data without sharing it leads to more value;
- The user utilizing the data themself leads to more meaningful action.
Let’s examine these three categories separately.
1. Continuous Data Availability Is Required to Generate Real-Time Action
Utilizing data for immediate action requires local data for almost all use cases. Being able to alert help when an individual has fallen or experiences a seizure would be time-sensitive and also required to function without an external connection, such as an internet connection. A self-driving car should be able to brake when its sensors tell it’s a local algorithm to, rather than a server externally. Latency or the lack of connection is a critical blocker.
2. Using Data Without Sharing It Leads to More Value
Some use cases imply different privacy levels for users to ultimately opt-in. For example, an employer-employee relationship is often such that if an employer provides a certain type of application, the employee may be reluctant to use it if their data is flowing directly to the employer without any type of filtering or user (employee) control.
We see applications that deal with workplace productivity, fatigue, and overwork providing a lot of value for the overall workplace and employees. However, employees, labor and data privacy regulations often prohibit applications that funnel all data about employees to employers.
Similarly, use cases where a company may be unable to use data points may not preclude the individual user from being able to themselves be able to utilize their data.
3. The User Utilizing the Data Themself Leads to More Meaningful Action
We are working with various partners utilizing wearable technology and data, and one key takeaway for our conversations has been that simply telling the user what to do will likely anger them and not lead to any meaningful change in behavior. For example, can you imagine your Fitbit device telling you to go to sleep — would that be the best way to get you in fact, to turn in?
Rather than sending data to a central server to crunch, simply keeping the data with the user, as well as the processing of that data, can give the user the power to make an informed decision. An example would be to tell users that when they have kept a regular sleeping schedule, their stress levels have fallen. Then leave it to the user to make the decision themselves. No data leaves the user, but the user can choose the action themselves.
While a simple example, similar types of logic can be applied to applications that monitor healthy behavior in wellbeing, work/overwork balance, and insurance applications. Where an incentive for the end-user to make optimal choices may align with the institutional stakeholder (e.g., employer, insurer) the ability to allow the user to make the choice themselves, also is complemented by what we covered in the 2nd section, data sharing to an employer or insurer is often inhibitive to real-world usage.
Therefore, allowing the individual to not only make an informed decision, but to retain the data and process the action on their side, delivers far more power for all stakeholders.
Diagnostic Tools for Our Bodies as Cars
I’ve used the car analogy once or twice for edge processing; so let’s use it for one more. Our cars have a lot of highly advanced analytics tools to let you know when they need service, what the tire pressure in PSI is in each tire, and what actions to take.
Personal AI is inevitable to fill this gap: if and where individuals can collect and utilize their own data, they may be able to unlock far more agency from their health and wellness and make more informed decisions to live healthier or more active lives.
Whatever their goal, it should be their own choice and any applications should empower them to make informed decisions.
Connect With Us and Stay in Touch
Prifina is building resources for developers to help create new apps that run on top of user-held data. No back-end is needed. Individual users can connect their data sources to their personal data cloud and get everyday value from their data.
Follow us on Twitter, Medium, LinkedIn, and Facebook, or listen to our podcast. Join our Facebook group Liberty. Equality. Data. where we share notes about Prifina’s progress. You can also explore our Github channel and join us at Slack.
