Wherever computing systems go, security threats soon follow. Microsoft released Windows in 1985 — and the first Windows virus, Brain, swiftly followed in 1986.
As edge computing matures, it faces similar cybersecurity challenges. IoT devices are vulnerable to threats such as botnets. Early fitness trackers were quickly hacked back in 2015. With the current reliance on the cloud and the increasing accessibility of powerful, smart devices, edge computing is the future for many industries. So we're bound to see similar attacks materialize.
Edge computing is now a $16.45 billion industry, and rapid growth could lead to annual revenue of $155.9 billion by 2030. As we explore the advantages of edge computing, we must also keep security at the forefront.
What Is Edge Security?
Edge security is a form of decentralized security that occurs at the "edge" of the network — such as near or on end-user devices. While nuanced, edge security may either refer to the systems used to secure an edge network or security systems on the edge of a network.
As with many emergent technologies, edge computing covers a relatively broad spectrum. Edge computing, and edge security, may either occur on the devices that are closer to the end-user device (such as routers or localized data centers) or on the end-user device itself (such as a sensor or other IoT device). Regardless, computation closer to the device rather than in the cloud improves latency.
Not all security must be pushed to the edge, nor does a system generally rely on edge security alone. Consider a fingerprint sensor used to authenticate your access to your company's intranet applications. The fingerprint sensor on your phone or your computer will do the bulk of the processing, ensuring that your fingerprint matches the signature on your device. If your fingerprint matches, a successful authentication is sent to the access control server.
But the server can't just rely on an "OK" from your device. If you had hacked the fingerprint sensor, for instance, a successful authentication would be passed with flawed credentials. So, while the sensor does authenticate the user, the intranet server authenticates the sensor. In this case, the intranet server will ensure that the sensor can provide an encrypted handshake verifying that the sensor is tamper-free.
Edge Security Risks
Edge computing involves a distributed system, and any distributed system has a broad attack surface — a larger number of devices need to be secured. Each individual Internet of Things (IoT) device could be vulnerable and potentially render the entire network vulnerable.
When an organization computes on "the edge," the cloud server cedes complete control over data and processing. Consequently, it becomes harder to validate that data. In our previous example, the fingerprint data is scanned, analyzed, and processed by the sensor device, and a "Yes/No" authentication (along with the sensor's own validation) is passed along. The company intranet never sees the original fingerprint scans nor the analysis that it generates.
Further, because edge computing frequently involves user-controlled devices, it becomes difficult for IT administrators and cloud providers to control the status of those devices. Users may not patch their devices in time, may attempt to jailbreak them, or may simply use them in ways that could compromise them (such as by connecting them to an unsecured network).
And even when these devices are controlled, there are many of them; administrators may need to track hundreds or thousands of sensors and devices, complicating network security.
At a higher level, edge computing is frequently used for applications that demand high-level processing: self-driving cars, merchandise warehouses, and even healthcare systems.
Autonomous vehicles must analyze and react to their surroundings quickly. Here, the risk could be fundamentally one of life and death.
At a recent hacking conference, hackers compromised a Tesla Model 3 through its infotainment system, gaining access to multiple subsystems within the vehicle, such as the ability to control the car's doors and front hood. If such vulnerabilities aren't addressed, they become known quite quickly to the community, causing further security risks.
Edge Security Considerations
Securing an edge system begins by determining what data needs to live on the edge endpoints and what data needs to live on the cloud. Major considerations include the sensitivity of the data involved, the cloud computing burden, and the security of the hardware.
Data Sensitivity
If data isn't processed on the edge, it has to be transmitted. For industries such as healthcare, it may be safer to process sensitive data at the edge rather than risk transmission.
The healthcare industry is experimenting with remote patient monitoring, such as keeping track of glucose levels. An advanced edge computing system could monitor patients and send up appropriate alerts. A system without edge computing would need to send all the data to be processed to the cloud, creating opportunities for a data breach that would put healthcare providers out of compliance with data privacy.
Furthermore, there's an issue of connectivity; a healthcare system can't stop processing patient-related, essential data in real time just because connectivity goes down. An edge device will still be available if the internet goes out, but a cloud server will not.
Conversely, a fitness tracker doesn't store detailed patient information — and fitness trackers generally send data, such as sleep data, to the cloud to be processed and analyzed.
Computing Burden
If the data demands extremely high processing volumes, the burden may need to be shifted from the edge to the cloud. As an example, most "voice analysis" systems today (such as transcription services) will send voice snippets to a cloud server to be analyzed, as the device itself may not have the processing power — so the edge isn't a good use case for these.
Hardware Security
If the hardware involved is vulnerable, edge computing becomes a substantially higher risk. Edge computing for high-security data is generally completed on close proximity servers or devices rather than on an end-user device — for example, authentication for kiosk tablets within a financial institution would occur on the company's local authentication servers or WAN rather than on an individual's tablet or the cloud.
Edge Security Best Practices
In practice, edge security is simple: know who you are dealing with (authentication) and provide authenticated users with the least possible potential for damage (least privilege). To keep your systems secure, you'll need to follow a few best practices.
Automated, Intelligent Monitoring
Edge computing systems can become large and unwieldy. Automated, intelligent monitoring systems increase the chances of detecting security risks without demanding the attention of a human operator. Imagine a warehouse that has thousands upon thousands of sensors. An artificial intelligence-driven security system would validate that these sensors are fully patched, secured, and operating as they should.
Multi-Factor Authentication
Authenticating based on a single factor is too risky; authenticating with multiple factors reduces the chances that authentication could be compromised. Sometimes, this is invisible to the end user. For today's authentication sensors, edge computing devices don't just look at the authentication type — such as a fingerprint — but where the device was used and if the device is being used in an unusual context, such as outside of regular office hours.Â
Enforce Security Standards
Edge devices must be held to the same security standards as internal network services, and security standards must be baked into the initial network architecture and enforced from day one. If a server is zero-trust, the devices that connect to it should also be zero-trust — frequently, this becomes difficult because the user is in control. In an autonomous car, a user might try to disable some security features. But if those features are disabled, the car should not be able to operate or connect to the broader network.
Security-First Solutions
Purchase hardware and software that puts security first — identify companies that have not faced serious security breaches, have solid security plans, and have invested in security technology. Few companies build their edge computing systems from scratch. Rather, they go through a vendor that produces devices within their sector, such as a healthcare provider purchasing MRI machines. If the vendor has been the victim of malware previously, it may indicate they aren't prioritizing their security measures.
Conclusion
We rely on the cloud, and our devices are becoming smarter every day. We trust these devices with sensitive data across a multitude of use cases, and we frequently rely on cloud security or cloud services to protect us. Edge computing is fast becoming ubiquitous, as are those who seek to compromise these systems. A deeper understanding of edge computing is necessary to maintain the security of these systems.
To learn more about the advantages (and potential pitfalls) of edge computing, read Edge Computing vs. Cloud Computing, Multi-Access Edge Computing, and the Top 5 Edge Computing Benefits.