Urban Surveillance in Nairobi

Last November, the Metro Development Ministry launched the Nairobi Integrated Urban Surveillance System (IUSS) project. I’ve only recently taken a look at the RFP to get an idea of the solution’s design requirements and expectations. It’s a great start but falls short in one key area which I’ll touch on below.

First a little about the project. The idea is to add visual presence to Nairobi’s streets for safety and security purposes – particularly the crime prone areas – and to make traffic flow management smarter. The deployment will be carried out in (at least) two phases – the first involves design and implementation of a Video Surveillance solution and a Traffic Management solution, while the second adds on more “intelligence” to the platform through Facial Recognition and License Plate Recognition.

So there’s lot’s of process and technology covered here, and many more design considerations. Having been involved in developing solution architectures for large-scale enterprise IP video and urban security deployments, I know this project will not be a cake-walk. From the onset, the solution will be required to scale from 51 cameras (size of the “pilot-in-production”) to 25,000 cameras in the future so that requirement has to be built in to the original design. To ensure the quality of video is not degraded during transit, the back-haul IP network infrastructure will need policies implemented to meet specific SLA requirements for packet loss, latency, jitter and end-to-end bandwidth. This is in addition to the usual staples – security, availability, extensibility. The location of the video management servers will be critical – too far from the camera endpoints increases risks associated with video loss and too close increases costs associated with the additional server nodes. Since this design appears to be highly distributed, the design of the storage area network should be taken under careful consideration as video preservation is a cornerstone of this solution – accounting for sizing, scaling and fault tolerance. Power considerations – especially in Nairobi (enough said). Lot’s more to think about. This is one of those projects that would make most architects giddy.

Let me temper that with this: if largely homogenous projects have inherent non-trivial risks – integration projects, such as this, definitely have a multiplicative risk factor because they’ve many moving parts that need orchestration. Yes, this is anecdotal but has held up pretty well so far and with more use-cases I run into. So I can only wish the system integrator that won the bid for this project all the best. (Aside: $5million for all phases of the project, while no small change, seems rather low for the level of effort to be expended. Then again I’m not a sales guy so I could be wrong).

What’s missing from the solution? An important component of any large-scale solution: an information management platform.

One thing the system operators will quickly realize once the solution is live is that they will get bombarded with alerts and events constantly being sent from the edge devices – information overload. Even just those 51 cameras could get overwhelming especially since they’re doing up to 100% motion-based recording. Sembuse 10,000? Unless the agency plans on augmenting headcount up into the hundreds, none of this will scale or be effective at preventing or mitigating crime, completely missing the point of the solution to begin with. You need a way to ingest all the alerts and process them intelligently in software based on established policies and standard operating procedures to establish any correlations, increase situational awareness and provide distilled, pertinent, actionable information with which to make informed decisions during incidents. There are some great software solutions out there that address this specific need.

Will this solution be effective in preventing crime or at least as a deterrent? Possibly. What’s true though, is that for the system to be effective, several structural changes in the way the government operates will need to be greatly improved. The eventual department that will be charged with managing the solution will need to work in tandem with other law enforcement agencies to respond to unfolding incidents.¬† Other information government information sources such as car registration data will also need to be integrated. Video evidence would also need to be made admissible in the Courts to aid in investigations and accountability. Then there are privacy concerns. The counter-argument here is that the cameras will be installed in public spaces and it will be apparent through signage that the area is under surveillance. Anyway, at the moment not too many seem concerned. In fact, maybe a little intrigued by this newfangled thing that will be watching them from up above.

Without a doubt this solution would grant the government agency, and indeed the system operators, a powerful set of tools. And in the words of Peter Parker: “With great power, comes great responsibility”.