Wifi has come a long way since its introduction in the late 1990's as a relatively low speed, short range network connectivity option. After a brief battle with some competing conecpts like HomeRF, the 802.11b Wifi standard gained steady adoption as the best way to connect your mobile devices (or, often mobile device, having plural devices in those days was uncommon). Now, Wifi is progressing from a way to send data through the air, to a way to detect the presence of objects based on their interactions with the wifi signal.
Let's look at how this emergining standard might affect the security industry, particularly as it relates to video analytics.
Knowing what is happening, or has happened, in a given area has been the basic purpose of video surveillance since the first rudimentary systems were deployed over half a century ago. As time and technology progressed, various methods to try and help indentify key portions of video were developed - PIRs attached to DVR inputs, motion detection, AI-based video analytics, and other sensing systems like radar or lidar to try and determine when someone or something of interest was in the scene.
Today, AI-based video analytics, using DNN models trained with security-specific images is more or less the state of the art for object detection in video scenes. These systems work well when key conditions are met, but they are almost always either a minimally integrated add-on, or a feature of premium cameras. Either way, the per-channel cost can often be quite high, making it hard to justify widespread deployment or usage on everyday systems. Alternatives, like the Avigilon Presence Sensor, have attempted to solve the problem using a sensor not based on a video stream, but these approaches still tend to be costly and have accuracy trade-offs. Part of the reason that costs are often still hundreds of dollars per unit is that these devices are focused on relatively niche applications directly tied to video surveillance, making it hard for them to get mainstream deployments and the cost advantages that come with high volume electronics.
The proposed 802.11bf Wifi standard may be the first technology we have seen that has the potential for wide-spread adoption and near universal availability. In short, 802.11bf works by measuring and analyzing the impact on RF propagation of a wifi signal that occurs from everyday objects. Even if you are not an RF engineer, most people who have used wifi have realized how the signal can be impacted by walls, furniture, or seemingly random things like where you are standing in a room. This is because the RF signals from your wifi-enabled devices are being absorbed and/or reflected by just about everything in the environment. Different objects affect the signal in different, but predictable, ways. By modeling how various objects affect wifi signals you can beging to determine what objects may be within the signal area, and what those objects are doing, by analyzing the tiny changes in signal strength and quality occuring as a natural reaction to those objects. Companies like Cognitive Systems have been promoting this technology for several years, and it was available in consumer devices as early as 2016 in the form of home security systems that used this concept as an intrusion detector for an alarm.
802.11bf can work with existing wifi radios, it is really just a signal analysis software package overlaid on existing FCC-certified hardware, making it easy to deploy. In many cases, existing routers and other wifi-enabled devices may be able to be upgraded to provide this capability with little to no cost.
The liklihood of 802.11bf to provide pinpoint precision, such as to build the equivalent of a line-crossing rule, is quite low, at least for the initial phases of the technology. Also, it would most likely not be able to provide granular person-counting data, or differentiate one person from another. While this may start to sound like it will not be very useful, most mainstream applications really do not need data that is more specific than "there are 2 people in this room", particularly when you compare the cost of near-free for 802.11bf to hundreds or thousands of dollars per channel for an AI-based solution.
Where is 802.11bf most likely to be useful in the video surveillance space? Here are a few examples where this technology can likely provide significant value and challenge traditional approaches like video analytics or motion detection:
Basic presence sensing in an area - provide real-time alerts of a person in an area or augment video searches to make it easier to find the segments of video when people or vehicles were present.
Device or object sensing - while video analytics can only analyze objects that are openly visible in a scene and large enough to meet the minimal size/pixel requirements of the analytics, 802.11bf can detect and analyze the impact on the wifi RF signals from devices that are purposefully concealed or simply mostly obscurred from view. While it is unlikely that this technology could be reliably used to detect concealed weapons, it will likely be able to identify cell phones being brought into a secure area for example.
Through-wall sensing - know when a person is approaching an area to generate a pre-alarm video clip, scale up resolution or bitrate, or move a PTZ. This could also be useful for privacy applications such as hospital rooms or restrooms to know when a person is in a room without needing to rely on a video feed. This could be extrapolated to a kind of loitering alert if a person appears to be in a room for a longer duration than should be expected.
Dynamic lighting control - Turn auxillary lights on to increase image quality when 1 or more people are in an area. This can help reduce light pollution issues or similar cases where there is a requirement or desire to keep areas dark, but not miss potential intrusions.
Further, these wifi sensing devices will likely be employed for other purposes like controlling room lighting or HVAC settings, making the prevalance of the technology, and the availability of the data something that becomes much more ubiquitous than video analytics or other presence sensing devices. This should enable the security applications to be near zero-cost add-ons, further driving adoption of wifi object sensing technology.
The main downside, for now, is that we are still probably 2-4 years away from 802.11bf being fully approved and implemented to the point that it can be considered a viable option for these kinds of use cases. Until then, existing options like video analytics and people-counting style devices will likely remain the primary methods for providing this kind of metadata to a surveillance system.