Interview with Leo Levit About Interoperability of Onvif Profiles

GIT SECURITY: Who are Onvif’s target groups?

Leo Levit: From our perspective, Onvif has three main groups that benefit from our work: Firstly, the device manufacturers and vendors of security cameras, sensors, radars, and VMS etc. Those are the ones who can use and implement the specifications that Onvif provides. The second group is system integrators: these are the ones who take the products from the previous group, put them together and build the systems for end users. For them, Onvif conformance is a matter of saving money, because through interoperability of providers, integration is accomplished. And the third group are end users: That’s the group that really benefits from their own investment in Onvif conformant products. They invest in something that we have designed for them to be future-proof. When they decide to go with another vendor later in the process, they are not locked into a specific ecosystem or technology.

What is new in terms of specifications?

Leo Levit: To facilitate those ecosystems, Onvif addressed several things over the past couple of years. For the device manufacturers and those who worked with the specifications, we moved our specifications onto open source. It is now open to work with and contribute, which boosted the pace of innovation. For example, once something wrong with a specification was discovered, the process to get it right is extremely short and that is why open source is a great benefit for the organizations involved. When it comes to the actual industry we released a couple of profiles: Profile M and Profile D.

Profile M supports analytics configuration and information query for metadata, as well as filtering and streaming of metadata. It has interfaces for generic object classification, and specified metadata for geolocation, vehicle, license plate, human face and human body. If conformant products have native support for features, such as media profile management, video streaming, adding images in metadata streams, event handling or rule configuration, then Profile M interfaces for those features must also be supported. And if conformant products support analytics for object counting (e. g. for people, vehicles), license plate recognition or facial recognition, or MQTT (Message Queuing Telemetry Transport) protocol used by IoT (Internet of Things) systems, then there should also be Profile M event handling interfaces for those functions.

Profile D is for access control peripherals: Card readers, locks etc. A Profile D peripheral device captures input credential identifiers and passes them onto a securely located Profile D client, such as an access control unit or management software. The client, which stores access rules, schedules and credentials, can then take the access decision and send a command back to the peripheral device to grant or deny access, display a message or request additional input such as a PIN code.

How much interconnection between profiles is there?

Leo Levit: The way we build our use cases is that profiles can work together and enable different use cases to come together. For instance, if we take Profile M and want to transfer metadata, we can imagine a situation where visitors are using their mobile phone with a QR code that they got from the company they are visiting and they are showing the QR code to the camera. The camera transfers the QR code using Profile M and responds back to the lock which is Profile D conformant. So, the whole use case is now facilitated by Onvif specifications.

Individually, each profile is very defined, but they are built in a way to work with each other giving them a higher interoperability.

What is the biggest struggle at the moment?

Leo Levit: There are some trends in the market on the rise, such as AI analytics. So we need to make sure to facilitate the connectivity when it comes to the models used by neural networks. And we have another trend that’s very visible here at the show in Essen: it’s the cloud. How do we work with the security in the cloud? Do we need a standardized way of communication within devices in the cloud or different cloud components? Those questions are quite difficult to answer! This is what we are trying to address at the moment and to discuss how we can help this industry to build and continue to build systems with interoperability in mind even if we are moving to a slightly different domain. Today, we are a device- and client-centric industry, but the whole industry is starting to move towards the cloud or adding cloud components to improve the flexibility of systems. So we need to really understand how we tackle that challenge.

How do Onvif members engage in the development and the processes?

Leo Levit: The whole organization is member-driven. In general, there are several companies that contribute more than others, but it is completely by choice. If there is an issue that a specific member has that they believe the whole industry will benefit from solving, they will bring it to Onvif and they will find a way to collaborate with other companies to create a specification or a profile. Different companies engage in different ways in different questions and it is all driven by their own agenda for the market. If, for example, they believe the cloud is absolutely the way to go forward, they are going to contribute to the developments.

What else is there within Onvif apart from specifications and profiles?

Leo Levit: Onvif is not only about writing specifications. We also host a big database with all the devices that are verified and tested that conform to the profiles. It is the accurate source of information, if you have questions about conformance for a specific product. It can be accessed freely on our website.

Another thing which is very important: The conformance process and what it means. It is a self-certification process where you verify your product against a specific set of Onvif-specified features.