Point of View

Open Architecture: what it means and how to make it work

3rd March 2023
By Kevin Davies, Global Director, Ports & Borders, Smiths Detection

Where are we today?

We are living in an age of increasing and evolving threats – from explosives and biological pathogens, to cyber-attacks. The technology to counter some of these threats is already in use: high penetration capability X-ray units using multiple screening angles, best-in-class image quality, object recognition algorithms, trace detection and networked alarm aggregation.

Increasingly, Customs services are also looking for and expecting lower False Alarm Rates (FAR), material specificity and a higher degree of assured detection for the items/cargo under inspection. Advancements in computed tomography technology offer a path to achieving this taking advantage of the increased accuracy of Diffraction X-Ray when determining material specificity, for example.

Additionally, the adoption of technology such as this, and therefore the increased accuracy and identification capabilities, into the “process flow” of inspections at a port or a border crossing is leading to increasingly automated screening solutions which not only advance screening procedures and detection capabilities, but also reduce the operator burden, create higher throughput, and maintain system uptime to help keep both people and property safer, without the need for large numbers of highly skilled and very costly security personnel.

The future: Open Architecture

This being said, what does the future look like for screening operations and their many suppliers? I believe the key change factor is Open Architecture (OA). It allows hardware, software and algorithms from different suppliers to be easily “plugged together” into one solution. The aim of OA is to make adding, upgrading and swapping components, or technologies, seamless. The future of Customs security will be driven by data informing risk-based decisions and using integrated sensors and devices from multiple providers. It will require open equipment interfaces and common data formats with an oversight mechanism to provide assurance on aspects such as technical standards, certification and liability.

Key drivers for OA adoption include the need to respond quickly to ever-changing threats; to leverage new and developing technologies such as artificial intelligence; and to produce detailed management information from an increasingly complex screening operation. It is perceived as a more flexible approach which would accelerate innovation and reduce time to market. It is already commonplace to integrate certain technology from different suppliers, and developing this further to incorporate software and algorithms is an exciting concept.

Smiths Detection, along with other vendors and governments, have been striving for an OA format for datasets relevant to High Energy Non-Intrusive Inspection systems. In recent years, globally, dozens of deployments using the WCO-sponsored unified X-ray file format[1] for non-intrusive inspection (NII) devices (UFF V2.0) have been achieved and this format is now widely adopted and often mandated. The next step towards closing the gap of true Open Architected Systems is by enhancing UFF V2.0 and widening its scope to include generic API for multiple asset access and integration to the specified standard. An example of this is inclusion of Radiation Portal Monitor data according to the American National Standards Institute (ANSI) 42.2 standard, but the breadth and depth of opportunity goes way beyond this, enabling true data integration and facilitating interoperability.

The competitive landscape is now fast evolving to the benefit of the industry. Concepts such as federated learning models, interoperability between competitors and geographies, learning gleaned from other industries, all mean collaboration between established and new players on centralized platforms. We are finally seeing a thriving, highly competitive marketplace which satisfies the requirements for data security and compliance, while simultaneously we are witnessing an exponential improvement in terms of performance and, therefore, our goal of a step change in operational efficiency.

With the transformation of Software Architectures away from monolithic to containerized structures, multiple solutions / approaches can be offered. These services take advantage of artificial intelligence and machine learning services solutions using “Software-as-a-Service” or “Platform-as-a-Service” models, and can be provided on-premises or in-cloud as needed, using and taking advantage of the latest in edge computing[2] capabilities, while maintaining cyber security.

Some areas where OA is a game changer

The use of Open Architecture is going to enable progress in a number of areas.

Detection capabilities

The first is detection capabilities. OA will advance detection capabilities through ever more accurate and effective algorithms. These algorithms are being developed from machine learning techniques, both by OEMs, as well as an endless supply of third-party developers. From start-ups to established players such as Google and Microsoft, this technology is expanding exponentially. Using such tools will without question mean step changes in the Customs capacity to fulfil its mission of promoting legitimate trade and safeguarding society.

Their value can be clearly seen in image analysis operations, where image analysts must match X-ray images of scanned cargo with original manifest reports while also looking for potential threats. Algorithms can automatically highlight only those X-ray images where suspicious items have been detected, such as cigarettes or even dangerous levels of radioactivity, speeding up the overall analysis process and supporting the secure movement of goods and free flow of trade. It is a new competitive landscape, and this is driving innovation and development.

Connectivity

The second is connectivity. Thanks to OA, competitors and partners are already linking new and legacy equipment and technology onto the same platform. It is relatively easy to link different brands of hardware, but we still need to explore and define standard interfaces, and this will come at a cost, with IT infrastructure generally requiring to be updated. With solutions already existing in other industries, there is no reason to have a bespoke security version and, in fact, a proprietary option could make solutions unworkable and unsupportable. Smiths Detection’s Universal Checkpoint Interface was created with this challenge in mind. It is open and available to allow the easy and reliable interoperability of scanners.

Operations

Thirdly, OA will enhance operations by increasing the ability to refine the process of selection on which cargo or bags are scanned using available NII techniques.

Centralized screening

Fourth, OA can facilitate the development of remote and centralized screening via a wide-area network (WAN), a connected collection of telecommunication networks distributed across a large geographic area spanning multiple cities, territories or nations so that the component networks can exchange data within the defined WAN group. WANs can facilitate the real-time sharing of images between different areas of a building or sites (or even countries and continents).

What are the remaining challenges to full adoption?

To ensure the result is versatile, enabling peak performing and highly secure screening solutions, there are some complex challenges to be addressed along the way. OA is an exciting, yet challenging, prospect for regulators. While there is a broad appetite to exploit the power of OA, there is a recognition that, while a new application may perform well and meet approval at a certain point in time, we need to ask how we assure this compliance over the lifetime of the system.

Beyond the assurance of performance, how do we ensure that third-party algorithms do not become invalidated if the OEM makes a slight change to the detector array due to the obsolescence of just one component? Equally, how does one assure the operator that a minor bug fix in the algorithm has not compromised the API or affected the cyber security of the interface and system?

How this is being solved, maybe inevitably, is for operators to have one entity as the single point of responsibility, whereby an integrator – often the OEM – assumes liability, guaranteeing that any application that it hosts on its platform will operate not just with the performance, but also with the compliance required by regulators, over the complete lifetime of the installation. This is not without precedent – just look at the phone in your pocket – every app on Google Play or the Apple Appstore has to go through testing before it will be hosted. Just as you do not expect your device to slow down when you open multiple applications, as we approach a truly Open Architecture in screening goods, we have to ensure this data ecosystem works faultlessly.

While great progress is being made with the integration standard of the dataset format, is it time to level the playing field with performance? For example, rather than mandate the Image Quality Indicators and penetration performances on an NII system, why not standardize those, along with the testing regime in which they are met, for example? Instead of prescribing image quality and penetration of X-rays, why not guide member states to use the International Electrotechnical Commission or ANSI performance standards when procuring?

A transformational period at our fingertips

OA certainly has the potential to transform security screening, and collaboration is at its heart; only together can we make ports and borders safer and more efficient. The underlying goal will always be to improve operational performance, security outcomes and passenger experience – OA is a new means to that end. Similarly, the aim at Smiths Detection is to help solve business problems and meet operational requirements, plus shorten and maximize the return-on-investment period for customers. We are committed to OA and will continue to work with all stakeholders to address the challenges and risks and move it forward towards execution.

More information
www.smithsdetection.com

[1] For more information on the UFF, see https://mag.wcoomd.org/magazine/wco-news-89/customs-and-industry-collaborate-to-develop-a-unified-file-format-for-non-intrusive-inspection-devices.

[2] Edge computing refers to processing, analysing and storing data closer to where it is generated, to enable rapid, near real-time analysis and response.