Our vision on the Security domain can be summarized in the following action lines:
- Use redundancy to allow actors to collaborate with and monitor each other;
Secure systems generally have redundancy built in to some extent to ensure sufficient uptime and eliminate single-point-of-failure. We envision taking this a step further; by setting up redundancy in such a fashion that two components overlap each other, they can monitor each other to ensure each component is still performing their assigned duty. A simple example would be two cameras in each other’s line of sight, so that if one camera becomes obstructed or otherwise faulty, the other camera can relay this information. Furthermore, individual components set up in this way can collaborate more effectively as well, such as two smoke detectors working in parallel to triangulate the source of a fire in a room.
- Ensure a continuous image of a situation and how this develops, using any available source of data;
Most systems are set up in a way where sensors relay information to a central control system, where interpretation is done by human actors. An example of this is a fire safety installation, which shows the status of a building using lights per room or zone. However, the system itself could also already perform an analysis of the situation to see how it will develop, such as tracking the direction a fire spreads through the measurements of multiple sensor. Finally, instead of relying merely on internal data sources, the system could also make use of external data, such as data from fire safety installations in adjacent buildings, or city disaster feeds on Twitter or other social media.
- Be critical of any source of data, ensure the validity of the content and source, and use multiple sources where possible;
As a logical result of the preceding action line, each source – whether internal or external – should be examined critically, and the information it provides should be validated by other sources where possible. We envision that a smart system would do so in a dynamic fashion; instead of sounding an alarm once there is a critical violation, sensors or human actors could already be mobilized when a single anomaly is detected, with the goal of identifying whether it should be a cause of alarm. This information then needs to be weighed against the potential cost of a false-positive (such as a building-wide evacuation alarm) or false-negative (such as a dangerous gas leak which goes undetected), to minimize costly mistakes based on uncertain predictions.