printer friendly version

The TCO of cloud surveillance

Cloud services in the surveillance industry are nothing new, with healthy growth rates in some countries. In Africa, the growth has not been as good due to bandwidth costs and limitations, but if the TCO (total cost of ownership) and benefits are as good as people say, would the cloud not be a better option, even if the bandwidth is more expensive?

SMART Security Solutions asked two successful, home-grown cloud surveillance operators for their take on the benefits of cloud surveillance to the local market. Does cloud do everything, or are there areas where onsite solutions are preferable? We spoke to Verifier and DeepAlert, each of which sent us combined responses; therefore, we will credit the answers to the companies, not individuals.

Starting out simply, we asked what the advantages or disadvantages of cloud-based surveillance are; where is it a ‘no-brainer’, and where would onsite solutions be better?

DeepAlert believes that, in general, it is more of a ‘horses for courses’ decision. In many cases, a hybrid architecture approach, selecting either cloud or on-premises solutions for various elements of the surveillance operation is more effective. In terms of video analytics, however, they say cloud-based solutions are far more cost-effective.

With cloud-based surveillance, it is important to determine the degree of usage, says Verifier. “Are we talking about using a cloud-based VMS system streaming 24/7, or are we talking about using the cloud for specific uses? As the questions are leaning more toward the scenario of using the cloud as the basis of all video analytics, our thoughts are below.”

Advantages:

• With cloud-based surveillance, you can reduce many of the costs associated with licensing monitoring systems and setup costs.

• The remote accessibility means it is easy to use from offsite locations and can even be leveraged for international monitoring.

• Cloud-based video analytics can be an added layer of AI, with additional services that can be bolted on. For example, cloud-based surveillance can add AI to analogue systems.

• Cloud-based systems can add a complimentary layer to counter blind spots in existing AI or areas in need of enhancement. Some AI systems have greater alarm trigger differentiation capabilities, so it makes sense to use a cloud-based AI system to add greater differentiation in strategic areas (such as busy areas or areas with previous false alarm/overactivity issues).

Disadvantages:

• Additional layers mean additional cost. Cloud-based storage/hosting for everything becomes costly quickly.

• Cloud-based surveillance, if used as the sole/primary analytics system, can also leave a site potentially vulnerable due to Internet issues or other service disruptions.

• Continuing on that note, cloud-based systems may be vulnerable due to not having local redundancy measures. If a client is only using the cloud for all of their video analytics, then if the cloud goes down, there may not be back-up onsite servers to take over.

Where it is a no-brainer:

• For effective licence plate recognition or facial recognition collaboration, having joint databases hosted on cloud-based servers is crucial. In this scenario, collaboration is key, and isolation is vulnerability.

When is onsite better:

• ‘Onsite’ does not have to mean on location. When comparing cloud vs onsite, the idea is that onsite equates to physical infrastructure (servers and the like) at the client’s property, but offsite monitoring can provide all of the necessary non-cloud-based backup redundancy measures while not being housed and operated at the client’s physical location.

• Large-scale enterprise sites often have hundreds of cameras, meaning that it quickly becomes impractical to have everything streaming to the cloud due to bandwidth constraints and the associated costs. Having a ‘local’ offsite surveillance security provider with physical infrastructure and servers, and independent video analytics capabilities can be a great go-between to reduce the costs of direct streaming, while utilising the benefits of the often specifically trained cloud-based AI analytic models.

• In the case of sites with potential security issues (such as nuclear power plants, high-risk/national key point sites, or even banking sites), the use of cloud-based surveillance is highly risk dependent. Sometimes onsite, on-location security operations and management is the only option.

• On that note, cloud-based servers may have specific cybersecurity risks that are not associated with onsite systems. The ability to self-isolate with local AI-driven video analytics systems may be beneficial in case of an emergency or a cybersecurity breach.

Where does the TCO come from?

Acknowledging that TCO evaluates the complete cost of a system over its lifespan, say five years, what cost benefits do customers obtain over that period of time? They can pay a big fee upfront or a number of smaller amounts over a time period. What is the difference at the end of that period, and why would cloud be a better option (or not)?

According to Verifier, one of the major benefits of cloud-based surveillance systems is the ‘rental’ model vs the ‘ownership’ model. “With this, one can almost guarantee the software will be reliable and current/up to date and constantly self-training with potentially greater inputs. Without the need for costly upgrades or firmware updates, it may be more beneficial to use cloud-based surveillance.”

The ‘pay as you go’ model of ownership also means that you can access the technology needed conveniently, without any down-payment or installation and without the need for dedicated technical staff to address any updates or issues along the way.

While there is still a need for physical infrastructure, connectivity needs, and licencing requirements, these setup costs are often greatly reduced compared to the licencing costs of owning video analytics software. Verifier believes that partnering with an offsite monitoring company that owns advanced video analytics software and integrates with cloud-based systems, has an in-house dedicated tech support team, and can advise ways to ensure your systems operate effectively and that you invest in the appropriate technology from the start, may reduce your TCO compared to attempting to operate your surveillance system in-house using cloud-based systems.

In terms of video analytics, DeepAlert states that the ‘heavy lifting’ comes in the application of a deep neural network to analyse the image or video. “The heavy lifting requires significant resources (GPUs) to ensure accurate and fast computation. Replicating that computation scale on a site-by-site basis would be prohibitively expensive. Deep neural networks also perform better with regular system training using varied datasets. By utilising a cloud-based system that receives data from thousands of cameras across different scenes, the diversity of the data necessary for training is ensured. In essence, accurate, real-time, intruder detection is far more cost-effective when using a cloud-based solution.”

Is bandwidth still an issue?

One of the regular negatives brought up when talking cloud, is bandwidth. While there has been an explosion in fibre connections in many countries on the continent, the costs can still be high. LTE or 4G connections could also be used, but these costs are even worse. So what is the ‘bandwidth-friendly’ solution for cloud surveillance?

Verifier notes that bandwidth issues sometimes have more to do with the costs associated with data usage than connectivity. Having a fast connection but a small ‘fair usage’ policy or budget simply means a client may hit the end of their connection quicker.

Looking at Africa, the company notes there are many issues to consider when examining cloud surveillance, most in favour of some form of cloud operations:

• Collusion and corruption are major issues in SA (and some other countries). So, having a certain amount of information stored on the cloud can reduce the severity or likelihood of someone using your client’s information, or deliberately trying to crash the system.

• When considering the risks of governance issues and failed states, having cloud-based surveillance operated internationally means a higher degree of oversight and independence, while maintaining a high standard of service.

DeepAlert says its hybrid architecture (using basic onsite camera triggers combined with powerful cloud-based analysis) is very bandwidth efficient. There is also granular configuration both on the camera and the DeepAlert interface, which enables users to fine-tune the data flow, limiting the number of triggers through the use of regions of interest in the field of view and scheduling of alert rule sets.

One common solution to the bandwidth question is edge storage. Many service providers install storage systems onsite to record all the high-resolution video footage, only sending specific data to the cloud for analysis. Would this extra cost not reduce the benefits the cloud offers?

Verifier does not think so. “Backup and redundancy can go both ways; Onsite storage does not mean onsite AI software capabilities, and cloud-based systems may actually benefit from having local storage in the case of load shedding or for archival of CCTV footage. Using cloud surveillance systems often allows for the tech support and AI processing needed without having to have as many onsite staff to maintain the systems. There simply may not be the local expertise needed to operate surveillance systems effectively.

“We prefer not to think along the lines of either/or. Many of our clients’ sites use a hybrid solution to get the best of all types of surveillance systems. Having local storage but nobody to operate local systems effectively, or no local onsite storage, but only cloud reliance could just as easily negatively affect a client’s security.”

Hybrid options cover the bases

No single solution fits every need; therefore, it seems logical that each customer should choose the best option for their situation and budget. A location where security is critical cannot rely on the whims of Eskom or Internet outages, for example. A hybrid solution, making the most of onsite and cloud solutions would be a better answer to cover as many bases and potential risks to the security operation as possible.

Edge and cloud-based systems can be a great combination, says Verifier. In scenarios of poor or intermittent bandwidth, the edge devices can process and send pre-filtered/analysed alerts to the cloud/offsite monitoring provider at a greatly reduced volume than cameras streaming directly to the cloud, meaning there is less likelihood of a bandwidth bottleneck situation, allowing the cloud-based surveillance systems to work efficiently.

“The issue is cost and necessity. In some scenarios, you may not need a cloud-based system if you have an edge-based device. In small/medium residential or commercial sites, an edge-based device linked to an offsite monitoring service may negate the need for cloud analytics. A human can often do the post-analysis and verification of an alert as quickly and effectively as a cloud-based solution, which would ultimately need human verification in any case as no AI system is 100% accurate, especially when it comes to determining human behaviour. Some edge-based devices provide additional redundancy by providing outputs so that the AI can also trigger conventional alarms. For example, Verifier monitors edge, cloud and combinations of the two, with the client’s risk profile largely informing the technology requirements.”

Verifier continues, “Cloud-based systems are great for further analysis, where analysis beyond ‘is it a human or non-human’ alerts are concerned. LPR systems use cloud databases to access greater datasets for data matching and comparison. In this example, the cloud database is the main feature; an edge-based camera with advanced AI may be able to detect vehicles and license plates accurately, but without the real-time updating and collaborative management of cloud-based systems, they would have limited usefulness.”

Solutions on offer

In conclusion, we asked both companies what they offer clients in terms of cloud surveillance solutions.

DeepAlert offers AI video analytics to detect intruder threats in real time and reduce nuisance alerts by over 95%. This service integrates with various video management systems, such as Milestone, HikCentral, Listener, the Alerting interface, and DeepAlert Premium, and is cloud-based, enabling offsite monitoring companies to scale very quickly.

Verifier offers a range of services, from cloud services, for which it has built partnerships with several cloud-based service providers. Additionally, it provides clients access to its cloud servers for cost reduction, redundancy, efficiency, and support purposes. In terms of managed onsite solutions, it offers managed services to select clients who share its vision of centralising disparate systems and building resilient and efficient NOCs (which are essential to any enterprise in the IoT era). To accomplish all that, it also supports integration with numerous systems. Its central platform talks to security, BMS, and IoT systems, which are cloud- or edge-based, and it offers advanced automation, analytics, and reporting capabilities.

Share this article:

Further reading: