📷Types of wireless camera traps
Breaking down the strengths and weaknesses of different approaches to connecting camera traps to the cloud
Cellular cameras
Cellular cameras are the most common, cheapest, and easiest to install of all wireless camera traps. Most major camera manufacturers make cellular models, and Trailcampro.com does a fantastic job maintaining up-to-date ratings and reviews of what's commercially available (their customer support is also very knowledgeable and happy to help guide you towards making smart hardware decisions).
Typically, if you have decent cellular service in all the locations you want to put cameras (usually >50%, or over 3-bars, of AT&T or Verizon), we recommend using them. We do not recommend moving forward with cellular cameras without first testing service on the ground, everywhere you plan on deploying, but a good place to start if you're curious whether there might be coverage in your area of interest (and it's in the US) is the FCC's national map of cell carrier coverage.
Pros:
Cheapest of all wireless camera trap types and most user-friendly
Each camera is autonomous and does not need be wirelessly linked to or maintain line-of-sight with any others
Cons:
Strictly limited to areas with decently strong cell service
Recurring data costs
Locally-networked, radio-based camera traps
Locally-networked, radio-based camera traps, such as those sold by Buckeye, Unmanned Wireless Systems, and Cuddelink, are less common and much more expensive than cellular cameras, but they can be deployed in locations that lack cell service.
In these systems, cameras on the ground are wirelessly linked together to form a radio "mesh" network: when an animal trips one, that camera takes a photo and broadcasts it out to the next-closest camera in the field, and the cameras repeat this pattern of receiving and then rebroadcasting the image up through the network until it reaches a centralized base station.
The base stations typically consist of a radio receiver, which demodulates the wireless RF signal into a digital image, and a field computer, which either stores the image locally if it doesn't have internet access, or uploads it to the cloud if it does (see note about backhaul options below).
Backhauls
In IoT parlance, the backhaul is the method by which a local network is connected to the broader internet. If you are operating in a location that already has some points of internet connectivity (e.g. maybe a comms. tower or field station), placing your base station at one of those locations and leveraging the existing internet will be the cheapest way to get your data to the cloud. If not, you have a couple options:
use a cellular-connected base station (requires that there's at least one point of strong cell service in the vicinity)
install a high-bandwidth satellite connection (e.g. Starlink)
use the network "offline". Many use cases don't necessitate transmitting data to the cloud at all, but having camera trap images wirelessly transmitted to a centralized hub or base station in real-time is still valuable for reducing fieldwork
The key takeaway is that aside from the base stations - which serve as the gateways to the internet - none of the other cameras or devices in the networks require any WiFi or cell service - again, they communicate 100% through radio. So this, combined with the fact that you can daisy-chain them together to extend their range and circumvent challenging topography, is what allows users to capture real-time imagery across vast, remote locations with little-to-no internet access.
Line of sight
On Santa Cruz Island, we have had a lot of success using Buckeye X80 cameras and repeaters to transmit images over long distances (6-7+ miles between camera/repeater "nodes") across extremely mountainous topography. However, one of the major drawbacks is that to communicate with one another every camera in the network needs to maintain line-of-sight (LoS) to at least one other camera or repeater in the network. Large landmasses like bluffs and ridgelines will likely block signal and LoS, and dense tree-canopy and foliage can also attenuate and block the signal. In our experience on Santa Cruz Island, we've found that strategically placed repeater nodes can help circumvent just about any LoS challenge, but it should be noted that the need for repeaters adds additional cost to the network and complexity to the deployment. When evaluating LoS and the need for repeaters, we recommend conducting a viewshed analysis using Google Earth Pro (for desktop)'s viewshed functionality or something similar. See our guide for planning deployments for more information.
Pros:
Local networks can be deployed across vast, remote locations that lack cell service
Operational and data costs are low if you can piggy-back off an existing point of internet connectivity in the vicinity
Cons:
Significant up-front investment (cameras themselves are expensive and you may need additional hardware such as base stations and repeaters)
Cameras need to maintain line-of-sight with at least one other camera/repeater to connect with the network
Often utilizes proprietary radio technology for local networking
Satellite camera traps
Satellite-connected camera traps are a bit more of a nascent technology and less common than the other types (to the best of our knowledge no major commercial camera trap maker offers them off-the-shelf), but there are a handful of conservation groups and hardware developers working on bringing these systems to the market. The most sophisticated and general purpose is probably Conservation X Labs' Sentinel, but there are a number of other satellite-connected camera traps that were designed for narrower use-cases such as poaching or biosecurity.
The following description is a bit of a generalization because there are nuances to each system, but at a high-level there are a few things to be aware of when considering satellite-connected camera traps.
Because satellite-connected camera traps can autonomously transmit data from each individual device directly to the cloud and don't require the cameras to be locally networked (although sometimes they are), they offer similar on-the-ground simplicity to cellular cameras, but they can be deployed anywhere, not just in areas with strong cell-service.
At first glance this makes them highly practical and appealing, but there are some important limitations to be aware of. Again, these limitations may not apply to all systems, so it's important to do your homework and evaluate each product on it's own merits.
The most significant limitation that may surprise users is that most satellite cameras do not transmit images to the cloud/end-users, they transmit text that describes the images. That is not 100% true across the board - e.g. some cameras will transmit some images but not all - but in order to make using the devices cost effective at scale they typically rely on low-bandwidth satellite services such as Swarm, which is designed for transmitting small packets of text & numerical data from IoT devices in the field (think temperature and humidity from sensors and weather stations) rather than large image files. In short, they are highly bandwidth-constrained.
That leads us to another limitation that users should be aware of: in order to generate text descriptions of what is in an image (i.e. "a rhinoceros was detected at camera 5 at 3:15 on 1/1/23") the cameras use on-board artificial intelligence, or "edge detection", to guess what's in them. And because humans have little-to-no ability to look at the images themselves to verify that the AI was correct, that means you need to really trust the AI models will detect what you're hoping to see, which, in the case of camera trap images, are notoriously error-prone.
For use-cases in which a false-negative or false-positive might have costly implications, not being able to see the images remotely might make this technology a bad fit. However, there are use-cases that have higher tolerances for AI inaccuracy, or situations in which you can compensate for the uncertainty with more cameras. For example, if you have a grid of satellite-connected cameras on the ground and many of them start reporting wildebeest detentions all at once, there's a very good chance that there's a heard of wildebeest moving through your area of study. Or perhaps you're conducting an invasive species eradication and you're interested in long-term population dynamics - in that case, textual data with some level of inaccuracy might be just fine for charting the decline in detentions of the animal over time.
Pros:
Satellite-connected cameras do not need to be locally networked with one another
Can be deployed anywhere including areas without cellular or internet access
Cons:
Satellite-connected cameras with edge detection likely transmit text descriptions of the images rather than the images themselves
Requires a high degree of confidence in the AI models that are generating the text
Recurring data costs
Last updated