We live in a world where more businesses want actionable, real-time and up-to-date insights from their data as soon as possible. In such a reality, the traditional way of collecting data from various IoT devices and sensors, sending it to a centralised data lake, warehouse or the cloud to be analysed, and get insights can’t sustain the exponential demand. Of course, many organisations rely on this centralised way of analytics, and perhaps are doing fine if they operate in areas where data doesn’t change frequently. However, what if the data does change frequently? Certain industries such as oil & gas, energy, healthcare, financial services, retail, transport, advertising, in which small shifts in data can mean large scale safety, health or financial consequences, require real-time data analytics. Stepping to the scene: Edge Analytics (otherwise knows as IoT Edge Analytics).
What is Edge Analytics?
Thanks to the growing popularity and evolution of IoT, data processing and analytics have shifted to the edge devices providing immediate insights to control downstream systems, introducing Edge Analytics.
It is an analytics method of collection, processing, and analysis of data at non-central components of the system, such as sensors, network switches and other connected devices – edge devices. This way, insights are gained on-site closer to the devices where the data is collected, within the same physical environment that generated the data, rather than relying on a central location far away.
As it’s related to industries that typically use IoT devices to collect data, it’s also referred to as IoT Edge Analytics, typically associated with oil rigs, mines and factories which operate in low bandwidth, low latency environments. But, IoT is also utilised across various other industries where data collection is the most efficient way of leveraging technology for multiple business needs, thanks to the ability of IoT devices to generate a huge amount of decentralised data, which is especially applicable to large-scale industries operating in low-bandwidth environments.
Why it’s important, especially in 2021
We’ve covered what Edge Analytics or IoT Edge Analytics is, but let’s dig deeper into why it’s coming back to the hot topics shelf in 2021.
During the last year and a half, the world was forced to hypersonic digitalisation, completely changing our ways of working, socialising, production, entertainment, and supply chain function. All this puts an extra burden on the internet infrastructure and cloud computing to satisfy increasing demands for bandwidth and growing data volumes.
Yet, as we are recovering from the COVID pandemic, digitalisation efforts will no doubt accelerate. The pandemic exposed broadband limitations and pushed internet infrastructure to its capacity limit. And it’s going to continue on the same trajectory. Projections estimate that by 2025 there will be 20 million IoT devices and 1.7MB of data created per second per person.
Sending all this data to the cloud for storage and processing is slow, expensive and wasteful, making projects expensive and unsustainable.
Instead, working with the data locally, on the edge, where it was produced and is used, is more efficient than sending everything to the cloud and back. Edge analytics reduces latency, cloud usage and costs, independence from a network connection, and enables more secure data and heightened data privacy and even reduces CO2.
Tech giants have also recognised the importance of edge analytics and have made significant investments in the area. For example, Apple has acquired an edge focused AI start-up Xnor.ai, corresponding to their plans to run deep learning analytics models on edge devices such as phones, IoT devices, cameras, drones, and embedded CPUs. Also, both Google Cloud and AWS have included IoT focused products in their portfolios.
Additionally, as businesses are collecting and processing more data, the need for stronger data privacy and protection also grows, especially considering there are more data regulations and laws coming into force in 2021. Edge Analytics helps tackle this challenge by keeping the data locally, ensuring clear data ownership and reducing the risk of the data being attacked.
Edge Analytics vs Edge Computing vs Cloud Analytics
Compared to regular centralised analytics, Edge Analytics has similar capabilities, except that it’s performed on edge devices, which can have memory, processing power or communication limitations. However, edge analytics applications are tuned to work in these limiting conditions.
Edge analytics may sometimes be confused with edge computing and cloud analytics. However, it differs in some key features from the other technologies.
Edge Analytics vs Edge Computing
Edge computing is a method of data collection and data processing performed near the location where the data is either being created or consumed.
On the other hand, edge analytics uses these same devices and the data that they have already produced to perform deeper analysis and enable creating actionable insights directly on the device, describes TechTarget.
Edge Analytics vs Cloud Analytics
The main difference between these two types of analytics is that cloud analytics requires raw data to be transmitted to the cloud for analysis.
Both cloud analytics and edge analytics have their applications. However, edge analytics has two main advantages over cloud analytics:
1. Edge analytics incurs far lower latency than cloud analytics because data is analyzed on-site – often within the device itself, in real-time, as the data is created.
2. It doesn’t require network connectivity to the cloud, which means that edge analytics can be performed in bandwidth-constrained environments or places that don’t have cloud connectivity at all.
Edge Analytics architecture and workflow
The most common use of Edge Analytics is monitoring edge devices, particularly when there’s a need to monitor a large number of IoT devices. The data analytics platform analyses if the devices function properly. In a case when there’s an issue, the edge analytics platform can automatically make corrections. But if the issue is more serious and it can’t be fixed with automatic action, the platform informs IT people and provides actionable insights on how to fix it.
AImultiple has broken down the particular workflow steps of edge analytics tools:
- Sensors or devices at the edge collect data.
- Analytics capabilities within the devices enable performing analysis at the edge.
- If the device needs to take action, it does so, relying on the results of the analysis.
- Relevant data is transmitted from the edge to the cloud so businesses can see the big picture by aggregating summarized data (in case of bandwidth constraints) from thousands of devices.
As for the edge architecture, it consists of different types of devices depending on their role in a smart environment, described in detail by Sciforce:
Edge Devices are general-purpose devices that run full-fledged operating systems, such as Linux or Android, and are often battery-powered.
The most commonly known edge or smart devices are smartphones and tablets, but other smart devices within IoT edge architecture include microcontrollers like Arduinos and single-board computers like the Raspberry Pi.
These edge devices run the Edge intelligence, meaning they run computation on data they receive from sensors and send commands to actuators. They may be connected to the Cloud either directly or through the mediation of an Edge Gateway.
Edge Gateways also run full-fledged operating systems, but as a rule, they have an unconstrained power supply, more CPU power, memory and storage. Therefore, they can act as intermediaries between the Cloud and Edge Devices and offer additional location management services.
Both types of devices forward selected subsets of raw or pre-processed IoT data to services running in the Cloud, including storage services, machine learning or analytics services. They receive commands from the Cloud, such as configurations, data queries, or machine learning models.
Edge Sensors and Actuators are special-purpose devices connected to Edge Devices or Gateways directly or via low-power radio technologies.
The most common use cases of edge analytics are:
– In retail, data from a range of sensors, like parking lot sensors, shopping cart tags and store cameras, can be leveraged for customer behaviour analysis. Retailers can analyse the data from these devices to create personalised products with the help of behavioural analysis and targeting.
– Remote monitoring and maintenance. Industries such as energy and manufacturing require instant response when a machine fails to work or needs maintenance. With edge analytics, organisations can identify signs of failure faster and take action before any bottleneck can arise within the system.
– Smart Surveillance: Businesses can use real-time intruder detection edge services for their security. By using raw images from security cameras, edge analytics can detect and track any suspicious activity.
As we’ve also seen above, the benefits of edge analytics are numerous, especially in organisations operating in low bandwidth, low latency environments.
Edge Analytics enables Faster, autonomous decision making since insights are identified at the data source, preventing latency. For example, in environments such as oil rigs, aircraft, CCTV cameras, remote manufacturing where decisions should be made in a split second, there may not be sufficient time to send data to the central data analytics environment and wait for the results to impact the decision on time. It’s more efficient to analyse data on the faulty equipment and immediately shut it up.
The growing number of data and devices increases the strain on central data analytics. Edge analytics enables to scale the processing and analytics capabilities by decentralising to the sites where the data is collected.
Reducing bandwidth usage
Transmitting data from edge devices to the central analytics platform also grows with the number of devices. Many remote locations don’t have the bandwidth to transmit the data and insights. Edge analytics reduces the work on backend servers and delivers analytics capabilities to these remote locations.
Since data is not stored or communicated, but it’s generated and analysed in a network of devices, it enables total control over the IP protecting data transmission. After all, it’s harder to bring down a whole network of devices instead of one centralised system.
Ultimately, edge analytics reduces the overall cost by minimizing bandwidth, scaling the operations and reducing the latency of critical decisions. But also, it lowers the cost of central data storage and management since less data is stored centrally and cost data transmission since less data is communicated to the central data warehouse.
When should you consider Edge Analytics?
Although Edge Analytics offers amazing capabilities, it should be seen as a replacement for central data analytics, KD Nuggets advises. Instead, both analytics can supplement each other in delivering data insights and both models have their place in organisations.
One limitation of edge analytics is that only a subset of data can be processed and analysed at the edge and only the results may be transmitted over the network back to central offices.
There is a tradeoff between loss of raw data that might never be stored or processed and latency. So when you consider Edge Analytics, it’s crucial that you strategically make the decision if having raw data stored is more important than latency. However, if latency is not acceptable, such as in time-critical decisions in remote energy or manufacturing sites, then edge analytics is the right method.