The biggest market sectors for the Internet of Things (IoT) are industrial, consumer, and Smart Cities. Of the three, Smart Cities face the greatest challenge when it comes to scaling affordably. Why? Simply, it is easier to connect devices and transport data within the defined perimeter of a building, campus, or home – and when there is a clearly defined hierarchy of ownership.
Cities are different. A municipality needs to connect nodes and sensors across a broad landscape, much of which is public right-of-way. This creates a lot of “gray areas” and questions for integrators:
- Who owns this utility pole, and do I need a permit to mount a device on it?
- Is there hard-wired communications infrastructure in that location that I can use to backhaul data, or will I need a wireless solution?
- If relying on carrier-provided wireless connectivity, which carrier offers the best coverage and affordable data plan? Will the plan provide enough speed and throughput to accommodate all the city’s sensors and application needs?
- Whose budgets can be tapped to pay for all of this?
Smart Cities have the potential to grow like enterprise deployments on steroids. Instead of hundreds or thousands of connected endpoints, a Chief Information Officer might be looking at tens of thousands of devices. Many of those devices will be high-resolution IP cameras that effectively distribute data to operating systems throughout the city. This is not run-of-the-mill city surveillance deployments that are singularly provisioned to improve public safety; instead these cities are using IP cameras as the ultimate sensor, delivering data to all branches of the local government to assist in myriad goals, including:
Crime prevention and event investigation; traffic optimization and signal control; emergency notification on weather, environmental conditions, and other immediate threats to the community; parking management; and much more.
The promise of municipal IoT is great, but the challenge is managing the massive data load without compromising video integrity, and without overwhelming bandwidth, storage, and budget resources.
There are three developments that give Smart City managers optimism that they can overcome these challenges: Innovations in compression technology; new wireless communication strategies; and emerging edge-based processing.
These touchstones represent a few of the innovations that can help municipalities become more efficient in how they use technology to improve the quality of life for their communities, as sensor and platform manufacturers, integrators, and service providers are becoming motivated to deliver solutions to satisfy the unique needs of the market.
Compression Technology: Smaller Packages
Historically, the security industry used video compression formats developed by the broadcast industry; however, those formats did not address the unique challenges faced by surveillance cameras – the need to capture forensic-quality video in constantly changing weather and lighting conditions, 24 hours a day.
In 2015, video manufacturers began tweaking the broadcast industry’s H.264 compression format with compression methods purposefully developed for surveillance cameras. They dealt with issues like dynamic conditions, long periods of inactivity, and grainy video generated in scenes with poor lighting. These compression technologies were developed specifically by camera manufacturers and their performance results varied greatly. The most broadly accepted formats significantly reduced the bitrate of video streams while simultaneously preserving image useability and forensic detail. Equally important, they maintained compatibility with H.264 standard decoding resources in centralized servers and video management software.
At the same time, the industry was also improving low-light sensitivity in cameras, which in turn reduced video noise in low-light conditions – a common cause of spikes in bitrate.
The improvement on these two fronts – video data compression and low light sensitivity – has not slowed in recent years; rather, they have accelerated to a point that camera technology today exceeds what was dreamed even possible less than a decade ago.
Moving beyond traditional use of surveillance cameras for physical security and towards using IP cameras as the ultimate sensor in Smart Cities, the benefits become even more pronounced. Effectively, the industry is getting better at providing more actionable information (forensic detail and metadata) with less data (smaller packet size, lower streamed bitrate).
But innovation is far from complete. Smart algorithms can now be layered onto H.265 to reduce bitrate even further; however, spikes still occur when capturing complex scenes with lots of activity. This makes it very difficult for a solutions architect or engineer to accurately anticipate how much data will be generated by any camera and streamed across the data communications infrastructure. How can one design a network or estimate centralized processing and storage requirements when they do not know how much data to expect or when to expect it? The answer lies in a new video data management technique: delivering video in an averaged format across a baseline developed at each camera.
How does this work? When the bitrate spikes, the camera will temporarily reduce the frame rate or resolution and hold the full video in cache for a time period, until the spike drops. Then the camera will push the held data to the central repository where it is associated with the video that was initially streamed.
This is an effective way to manage the massive data traffic of a Smart City, whether the devices are transmitting over a wireline, private wireless network, or carrier-provided wireless data pathway. It is important to note that this technique does not reduce the total data streamed, it simply evens out the volume of data streaming at any given time to ensure the infrastructure is not overwhelmed.
Wireless Networks: Metadata and Video on Demand
As cities deploy their surveillance cameras and IoT devices across their public rights-of-way, building out infrastructure for sourcing power and maintaining connectivity to remote locations becomes an issue. In lieu of trenching fiber everywhere, cities might consider relying on cellular networks to move data from sensors; however, that solution becomes prohibitively expensive when you attempt to scale it to a city-wide level because data plans are generally designed for the consumer subscriber market.
Carriers need high-density IoT deployments in the public right-of-way to justify building out the required small-cell network infrastructure necessary for emerging generations of wireless communications.
In the new business model, wireless carriers are starting to develop data plans better suited to the high-transmission traffic typical in a Smart City deployment. Even with a more robust cellular pipeline, however, continuously streaming live video would consume a great deal of bandwidth.
The new strategy is to configure the system to make real-time use of metadata from the camera, and to only stream video on-demand. For instance, when an operator receives an alert of a detected event, they could confirm the situation by pulling a snapshot or short video clip from the camera. The camera could store the full video record and upload it to the central core when other demands for bandwidth are low.
Edge-Based Processing
In the past, video analysis could only happen at the server level, which often delayed response times to critical events. Now that surveillance cameras have more powerful processors and artificial intelligence capabilities, complex analysis can take place at the edge.
Edge processing offers multiple advantages: It allows the cameras to select the relevant video to transmit, which helps the operator immediately focus on the key details. Reducing the video stream reduces bandwidth consumption and storage needs.
Just because the camera is not streaming video 24/7 does not mean it is not working for city stakeholders. Edge processing with deep learning enables a camera to generate rich metadata that classifies objects within the scene – such as pedestrians, cyclists, passenger vehicles, large trucks, etc. This data can help planners and civil engineers understand how people and products move through the environment. Armed with these insights, stakeholders can make basic decisions like where to add a bicycle lane or where to redesign an intersection to improve community safety, efficiency, and equitable distribution of city services.
Using cutting-edge computer science for aggregating rich datasets empowers Smart City stakeholders to make informed decisions more likely to elicit positive results that enrich the lives of those who reside, work, learn and play in their communities. Edge processing is critical for making Smart City solutions more scalable and resilient, while simultaneously lowering total cost of ownership, so that the vision of Smart Cities can be delivered to communities.
Kevin Taylor is Smart Cities Segment Development Manager for Axis Communications. Request more info about the company at www.securityinfowatch.com/10212966.
