printer friendly version

The edge of surveillance

The edge is in vogue and all the cool kids are doing it – and not only in the surveillance market. In the broader IoT (Internet of Things) world, the edge is where it’s at. Many companies are touting the benefits of doing your processing at the edge (as in at the edge of the network instead of on a server) as a way to increase efficiency and reduce costs.

A similar view is held in the surveillance market as can be seen in this year’s handbook, although there is still definitely a need for server-based processing – whether on one’s own infrastructure or via a cloud service. But how important is edge computing today for the many vendors that have made their reputations by providing server-based surveillance storage and analytics?

Quintin van den Berg from Bosch Building Technologies believes edge processing has been shaping the market into a very intelligent environment with new opportunities for individuals and businesses. He notes that there is a ‘but’ in all of this. “In order to take full advantage of this, the security and safety industry must be ready for change. This has been at the centre of Bosch’s thinking for many years and has driven us to launch a new open platform, INTEOX.

“Today’s security solutions have far more potential than simply creating a safe and secure environment. Security devices are actually sensors, capable of providing valuable data that can be analysed and used in new, beneficial ways for solutions and applications – for example, true business insights.

Danny Pringle, director of Secutel Technologies agrees, adding that edge processing will increasingly play a more significant role in future video-based technology platforms. As can already be seen, camera manufacturers are releasing more AI functions as standard features on their cameras at very little additional cost.

Danny Pringle.

However, he warns, “The mere fact that processing on the edge is possible, does not mean that a ‘one size fits all’ approach can be followed. To find the balance between what should be processed on the edge and what should be processed on the server, will be dictated by requirements and not by technology. It is also our opinion that, similar to what happened with ERP systems, on-demand and scalable cloud-based surveillance and analytics services will increase dramatically over the next few years.”

Edge processing is, however, inevitable. Pringle explains that the lack of bandwidth and unstable power availability in South Africa will escalate the adoption of edge processing.

The cyber question

Simply moving the processing to the edge, whether in the surveillance market or anywhere else, may sound like a good idea, but the question of cybersecurity must be addressed. There are many stories about edge devices (even a fish tank thermostat) being hacked and allowing criminals to gain access to networks and servers and hence sensitive information. Are the manufacturers and installers aware of the need for enhancing cybersecurity?

Since there are continuous advances in cyber-attacks to meet and mostly exceed what manufacturers do to secure their devices and services, Van Den Berg says that no manufacturer can underestimate the seriousness of cyber threats, regardless of how intelligent their edge devices are.

If the necessary actions and development are in place from a data security point of view, he notes that the systems will be relatively secure and have measures in place to send notification if they are breached. “It’s like any other security measure in the ITC/security realms, the better you keep the system updated and maintained, the better the product will perform and deliver.”

Another challenge is the high level of development of camera-based AI at the moment. Pringle says there will be constant changes and improvements to on-camera applications, which can result in vulnerabilities. “This will require the introduction of disciplines that the management of security technologies never had to concern themselves about in the past. This will place strain on infrastructure, operations and system integrity.”

As Pringle mentions, edge-based processing brings new challenges and opportunities for cyber criminals, but also adds the question of the operating system (OS) on the cameras being able to handle whatever is thrown at it. As the article on the S&ST; standardised camera OS explains, with rapid developments in AI, some users may want to recommission cameras with different intelligence as required. This sounds good, but is it a reality – whether using the new OS or other vendors’ firmware – when considering the vulnerabilities of ‘leaving a door open’, even if it is secure?

Moving to a new normal?

Van Den Berg believes that if this is to be achieved (along with more edge-based processing in general), “we will require a more open platform that combines built-in artificial intelligence, the right levels of performance and a more commonly-used operating system with the ability to freely add or remove software apps without boundaries. By offering this, we open up to a world of new developments and possibilities. It enables more opportunities for customisation and third-party innovation.

“It’s an approach that has proven very successful in the PC and mobile marketplace where many software and app developers have substantially increased the use and applications for consumers and the industry. The future growth of our industry will be accelerated if developers, system integrators, partners and other market players can independently imagine, create and deploy customer-specific solutions consisting of apps, software and services.”

This is not a simple switchover, however. Pringle believes that most organisations still have a substantial investment in cameras that have neither intelligence nor the capability to support the S&ST; OS. “It will take some time for this to become a reality. The organisation needs to have the financial and/or operational appetite to rip-and-replace their current technology base.”

This is a fair comment and the reason Bosch, for example, still sells a broad range of surveillance cameras with its own OS. However, the momentum has started and Bosch cameras based on its new open camera platform, INTEOX, follow the specifications set by the Open Security & Safety Alliance (the OSSA technology stack) and use the open OS.

But a new OS and ‘ripping and replacing’ is not necessary to capitalise on these technologies. Pringle notes that it is not necessary for cameras to have storage and AI capabilities to be able to experience the benefits of edge processing and AI applications. “The main driver for us at Secutel to develop our SecuVue solution was to enable organisations to retain their current infrastructure and add edge processing, edge storage, AI and selective exchange of data with servers in the cloud in a managed and controlled manner.”

What is out there now?

Pringle adds that the SecuVue solution is a scalable cloud-based platform, locally designed and manufactured (by Secutel) that uses the SecuVue Cloud Bridge or SecuVue Cloud Video Recorder. These are installed on the client’s premises as an edge processor.

“SecuVue is camera brand agnostic and as such, the client’s existing cameras, which need not have any intelligence or storage, are linked directly, or via an existing NVR/DVR, to the SecuVue device. The business rules associated with each camera are configured and maintained in the cloud.

“It makes efficient use of bandwidth by only sending relevant frames to the cloud for further processing. The user can then select an image from a photo catalogue and request high definition footage to be downloaded for local processing. Such footage has a secure watermark, which ensures the integrity of footage.”

As an example, he says people counting is performed onsite on the SecuVue edge processor while many-to-many facial collection and recognition, such as will be required in front of an ATM or at the entry of a bank branch or retail store, is done in the cloud and checked against the national database of persons of interest. Current analytics available using SecuVue include facial collection and recognition, automatic number plate recognition, people counting, line crossing, age estimation, emotion detection, gender detection, activity heatmaps and scene analysis and behaviour monitoring.

Expanding on the Bosch INTEOX camera range, Van Den Berg explains it offers the following:

• Built-in video analytics as standard at the edge (like all other Bosch IP cameras).

• Uses a standard operating system.

• Abundant developer libraries and codecs to support the development of apps.

• There is a large community to tackle possible vulnerabilities.

• Supported by the quality foundation Bosch cameras are built on.

• The INTEOX open camera platform is like other Bosch IP cameras, fully supported by the Bosch cloud infrastructure (Remote Portal).

The range is further supported by the Security and Safety Things IoT infrastructure with:

• A digital marketplace (application store).

• The developer web portal for the development of third-party apps.

• Device management portal for the purchase, installation and management of third-party apps.

Share this article:

Further reading: