Smart lighting is an integral part of the city of the future. It provides a safer feeling, a better urban experience and - not unimportantly - energy savings. Imec, the City of Antwerp, Digipolis Antwerp and Nokia Bell Labs joined forces to develop this kind of dynamically controlled street lighting. The project lasted two years and was tested in Antwerp with five potential scenario’s. Thanks to Nokia Bell Labs' worldwide streams (WWS) platform and imec's rule engine, among other things, it turned out to be possible to handle the large amount of data (from motion sensors and cameras) efficiently and to make decisions and control the street lighting without any significant delay. The combination of cloud and edge data processing proved to be a hit.
This article was originally published on the imec City of Things blog.
Koen Triangle, project manager imec City of Things and Lode Hoste, Senior Researcher at Nokia Bell Labs talk in more detail about the technologies used in the smart street lighting project.
Why would we need smart street lighting?
A concept such as smart lighting has a great many potential applications. For instance, imagine being able to have motion and sound sensors that are capable of responding flexibly to the start (and end) of your basketball game in such a way that the light level is only increased to its highest when the game is actually being played; or imagine your visibility and sense of safety on the street being increased at night because the lighting along your route can be activated automatically based on observations from motion sensors and cameras; or imagine the lampposts in your neighbourhood also acting as weather stations, enabling light signals or colour codes to warn you in advance of any approaching rain or snow showers; or imagine discouraging night time noise by automatically switching streetlights to a (brighter) emergency mode whenever the sound sensors detect suddenly louder noise levels; or imagine being able to switch on street lights automatically (with the appropriate light intensity) when the light sensors detect a reduced light level for some reason, also reducing energy consumption and light pollution.
All of these scenarios have been tested (and will continue to be tested until the end of December 2019) in the Antwerp Smart Zone in the framework of the smart lighting project and involve the rollout of two vitally important brand-new technologies.
What if smart street lighting would only work if the city's basketball court was actually used? It is one of the energy-efficient applications of smart lighting.
Nokia Bell Labs' worldwide streams (WWS) platform: solving the latency problem by combining cloud and edge data processing
The first technological part of this smart lighting project was developed by research partner Nokia Bell Labs. Using their WWS platform – a large-scale platform for processing multisite video and data – the information supplied by the numerous sensors and cameras located in the Sint-Andriesplein area can be processed in real-time.
In practical terms, this means that all sensor information can be processed and aggregated as quickly as possible to create once per second a representation that contains all of the highly semantic information enabling the correct actions to be carried out.
The subsequent control task required for actually controlling the smart street lighting is performed by the imec rule engine (but more about that in a moment).
Through the WWS platform, Nokia Bell Labs aim to take the next step in the smooth rollout of Internet of Things (IoT) services, including smart street lighting. At the moment, though, IoT applications, whose success depends mainly on real-time decisions, still have to contend with the restrictions of the underlying telecoms networks and their specific architecture.
One of these limitations is the issue surrounding latency. Latency is the time-lag between which signals are sent and received (one example is the delayed response from the person you are speaking to during a videoconference). The whole issue is bound by theoretical limits, such as the speed of light.
However, the latest generation of telecoms networks has already progressed significantly and as a result, their responsiveness is very close to that of the speed of light. So that is an area where we should not expect any new breakthroughs for the time being.
According to the researchers at Nokia Bell Labs, the key lies in a more distributed architecture – with cloud and edge components – with which the WWS platform is fully aligned.
“At the moment, (IoT) software is typically written for cloud environments,” says Lode Hoste from Nokia Bell Labs. “This means that you have to send all of the information from all of the sensors to the decision-making logic at the center of the network in order to reach intelligent decisions. As a result, the underlying network often finds itself overloaded, which in turn causes latency and generates higher costs, as you might expect.”
“The WWS platform on the other hand is built specifically to break the software into multiple building blocks, which means you can host some components closer to the source (the edge of the network), while others are handled more centrally in the network (the cloud).”
“In the case of the Antwerp Smart Zone for instance, where we use motion cameras to control the streetlights, this means that you do not have to send all of the images across the whole network. The image-processing itself – i.e. looking to see whether someone is approaching or not – is done as close as possible to the camera. This in turn results in us being able to reduce the volume of source information that we have to send to the central decision-making logic by no fewer than two orders of magnitude!”
“This way of working also delivers a great many advantages in terms of security and GDPR regulations,” adds Koen Triangle (imec City of Things). “In fact this local processing, close to the sensors, means that there is a clear delineation between where the images are generated and where they are processed – which provides the best guarantees when it comes to data protection.”
For this local processing, the Smart Zone partners are currently using specific hardware – such as imec’s very powerful DGXI server, which is hosted in the network of another partner, Digipolis. “But we also know that some cameras are already equipped for supporting direct local processing in the near future,” observes Lode Hoste.
“The WWS platform aims to be as open and accessible as possible and so has no specific need for sensors,” he concludes. “In the case of the Smart Zone cases, after an extensive market survey and various lab tests, we decided to opt for sensors that support the AMQP protocol, a popular brokering system for communicating real-time events and streams.”
The rule engine of imec
“The rule engine – developed by the researchers in the imec Application Prototyping Team – is the beating heart of the smart lighting project,” says Koen Triangle. “It contains all of the logic for the system: when a specific trigger comes in via the WWS platform, it is the rule engine that knows – and which decides – what needs to happen.
An example: if the motion and sound sensors on the basketball court detect that a game is about to begin, the rule engine receives this information in real-time via the WWS platform and uses it to decide, also in real-time, when to increase the light intensity of the court lighting.”
“That’s what is so unique about the whole system: the combination of the WWS platform and our innovative rule engine enables events to be processed in super-real-time, with the lowest possible latency – and it makes all the difference compared with existing systems.”
Also useful for other applications such as smart, green energy supply?
The smart lighting case in the Antwerp Smart Zone is coming to a conclusion now. The months ahead will mainly be about keeping the platform operational and gathering additional feedback. In the end, the project will conclude with a final report featuring a detailed analysis of the various cases and the main lessons learned.
Both the WWS platform and the rule engine have since demonstrated that they are broadly applicable and can also support a number of other cases.
“One of the other cases we have developed in the meantime is in the area of smart, green energy storage. This is a market with many consumers, as well as a growing number of producers (for example just about anyone who installs their own solar panels). Coordinating production and consumption in such a volatile environment is a difficult exercise – but it has now been modelled fully on the WWS platform,” says Lode Hoste.
The imec rule engine in this particular project has also been set up as a tool in which developers can define logical rules about real-time data flows (in this instance from the WWS platform). The next version of the rule engine (for which development is currently in the final stages) will make it possible to configure a large part of these real-time rules via a visual ‘drag-and-drop’ interface – without any knowledge of programming. With this visual representation, the logic control of an IoT or smart city system will become more insightful for stakeholders and enable them to make adjustments to controls where necessary.
Want to know more?
The smart lighting project came about as the result of collaboration between several project partners. The contribution of Nokia Bell Labs and imec is dealt with in detail in this article and special thanks also go to: City of Antwerp, which coordinated the installation of the smart lighting technology, including by keeping the platform open to control the street lighting, taking responsibility for interaction with Fluvius; Schréder, which supplied the lighting fittings and the lighting control platform. Schreder also provided intensive support with connecting the platforms; Digipolis played an advisory role, supporting how and if the project can be translated over time into permanent implementation.
Stay up to date with all the news about imec City of Things via the blog or one of the social media channels (see links on the blog page).