Cities need to get smart about IoT

Cities are moving to the edge

Much of the use case for smart cities involves protecting citizen safety. This could mean detecting a crime in progress, alerting first responders to a house fire or directing emergency services to a citizen having a heart attack. Seconds matter in all of these scenarios — which is why edge computing is such a priority. Here’s an example: Let’s say a city installs a sensor that monitors air quality, and one day it detects an anomalous reading. It sends its readings to a data center hundreds of miles away. That data center then has to spend time processing the data and then has to spend time sending it back. Meanwhile, the gas leak that the sensor detected has turned into a roaring fire.

In a scenario where the air quality sensor is equipped with on- or near-device computing power, that fire doesn’t happen — or at least when it does, first responders are already on their way. The device uses its onboard computing power to conclude that a leak is occurring and then notifies the relevant authorities directly, without relying on backhaul.

Cities have recognized the potential impact of edge computing and a number of smart city projects have begun to incorporate it in a number of ways, including:

• Water quality: In New York City, a complicated and aging water system supplies millions of people. A recent Legionella outbreak affected an NYC hospital, but by using IoT sensors in the water supply, municipal workers were able to flood the connected pipes with elevated levels of chlorine, preventing the outbreak from spreading through the water supply.
• Municipal buses: Cities are working to make buses safer by installing recording systems that identify potentially violent incidents. It’s very hard to stream video and audio from a moving vehicle, so the IoT surveillance systems on municipal buses are provisioned with computing power that allows them to recognize and report problems.
• Highway tolls: Last but not least, almost every commuter on a toll road has been the beneficiary of an edge-powered IoT system. Connectivity is hard to come by on a busy highway, so license plate cameras take photos and conduct optical character recognition locally, then transmit to a central private cloud at night when there’s less internet traffic.

Edge computing makes sense in issues where urgency is of the essence and in situations where good connectivity is difficult to find. Cities have a lot of these areas, and the use cases above are just scratching the surface in terms of where edge computing can be applied. The only question is support — where should cities begin to implement edge computing in order to maximize outcomes?

How should cities build an edge computing infrastructure?

In concept, building out an edge computing infrastructure for smart cities isn’t hugely different from building out other IT services. Cities must define what they wish to achieve via IoT, define their priorities and support them accordingly. For instance:

• What is the goal of a smart city IoT project? Do you want to reduce accidental deaths, stop crime, monitor the quality of the environment, reduce traffic or something else?
• How important is the project? If it goes offline, will peoples’ lives be in danger? If so, is the danger immediate or long term?
• For mission-critical activities where lives are on the line, cities need to design redundancies into their systems and have support teams on standby to fix them when they break.

For some IoT services, such as highway tolls, uptime doesn’t matter as much. Cities want to achieve smart fare collection, but commuters won’t notice or suffer unduly if the system goes down for a while — and in the meantime, the license plate camera is still snapping photos, ready to upload them once connectivity comes back. The same can’t be said for something like a water-quality monitor. Here, it’s essential to have local computing resources on hand so that if a network connection goes down, city administrators still have access to real-time intelligence and analysis.

Additionally, analyzing the chemicals in the water isn’t enough. That analysis has to be done as close to real time as possible to prevent widespread contamination. Then, combining that data with numbers around flow volume and pressure, you can model the potential spread of a contaminant to be certain that you’ve flushed the water supply thoroughly. It’s all about combining data together in the right way and on a platform that can cope with the demands of the data analysis.

Based on our examples, there are three important elements where cities should focus their investments in IoT infrastructure: the core, the cloud and the edge. Edge computing may be more expensive as it involves purchasing instrumentation with added computing power for onboard analytics, but it’s useful for high-priority, low-bandwidth scenarios. A city’s private cloud has a little more latency and a little more computing power. It’s harder to get workloads there, but it’s easier to process them once they arrive.

Lastly, the public cloud has the highest latency and the highest amount of computing power. Data centers in the public cloud may be located in a completely different region, but their processing power is functionally infinite. This mixture of capabilities suggests a framework for smart cities to prioritize their workloads and prioritize their investments.

For a smart city project to work, administrators need to identify where their IoT systems can prevent life-or-death scenarios. They then need to determine where these situations may involve a combination of congested bandwidth and heavy workloads, such as image recognition or speech analysis. In these scenarios, taking compute resources out of the cloud and placing them close to instrumentation can have an immediate positive effect on citizens’ health and well-being.

All IoT Agenda network contributors are responsible for the content and accuracy of their posts. Opinions are of the writers and do not necessarily convey the thoughts of IoT Agenda.