We know you hate the Internet of Things, but it’s saving megafauna from poachers


Sean Gallagher, Ars Technica

Date Published

See link for photo, video & maps.

For much of this decade, organizations seeking to protect wildlife have attempted to use emerging technology as a conservation tool, allowing small numbers of people to monitor and manage data from animals over a wide area. 

Nowhere is that effort more focused—and more desperate—than in the regions of Africa where illegal animal trade is threatening to wipe out endangered animals such as rhinos, elephants, pangolins, and lions. 

Here, several organizations are applying Internet of Things (IoT) technology to protect animals, providing rangers with data that helps them intercept poachers before they can get to their quarry.

Many conservation efforts elsewhere use IoT to try to track the location of animals, such as Vodafone’s IoT tagging of Scottish harbor seals and tracking of endangered dugongs in Philippines. 

But in Africa, the task of protecting rhinos is slightly different—it’s about tracking people, specifically the poachers who hunt down the rhinos for their tusks.

Rhinos, of course, aren’t unique in needing such intervention. Based on data from the Great Elephant Census (GEC), a continent-wide survey conducted by Microsoft cofounder Paul Allen’s Vulcan Inc., Africa’s savanna elephant population declined by 30 percent between 2007 and 2014 for instance. 

That’s a loss of 144,000 elephants. Current data shows the rate of decline of the elephant population is now eight percent per year, and ivory poachers are the main reason for that decline.

Things are even more bleak for rhinos. 

While the black rhino has bounced back from near extinction in 1995 (rebounding from a population of less than 2,500 to approximately 5,000 today), the Western Black Rhino was declared extinct in 2011. And the last male Northern White Rhino died this March in Sudan. 

Overall, rhinos in South Africa are being killed at a rate of more than three per day. If that continues, rhinos will be extinct in South Africa by 2025.

But networking and sensor technology, in combination with analytics, are now offering ways to better manage populations of these animals and intercept pending threats. 

By connecting sensors to the cloud (public or private) over low-power networks, these tech conservationists can offer essential intelligence on human activities near protected animals and help intercept poachers before they can do harm—both to the animals and the rangers guarding them.

Eyes and Ears and Clouds

For a few years now, The Lindbergh Foundation’s Air Shepherd program has used a combination of drones and data analytics in a South African trial to protect rhinos. That program has been expanded to elephant protection in Malawi and Zimbabwe with crowdfunding.

As Ars reported in 2015, Air Shepherd uses a combination of local intelligence and intelligent image processing from drone sensor data to build a model for what’s going on around and within the parks its crews protect. 

Based on animal movement patterns and the proximity of human threats, the drones are deployed to give rangers advance warning of poachers’ movements. But Air Shepard’s drones require trained pilots, and drones—while often an effective deterrent—can’t always be on station to detect poaching attempts.

So another effort at South Africa’s Welgevonden Game Reserve has tried a different sort of IoT approach, tracking the behavior of other herd animals (such as zebras, impalas and gazelles). These monitors act as “sentinels,” essentially watching for variations in animal movement in response to different potential threats. 

The Welgevonden experiment, based on a collaboration with Wageningen University in the Netherlands and IBM, tracks the movements of these collared herbivores by utilizing collars transmitting data via a 3G wireless network, and the effort inputs that data into an IBM Watson system in the cloud. 

The analytics of the Watson IoT platform are being trained to detect the difference in behavior in herd animals based on whether they are coming in contact with natural predators, tourists, or potential poachers.

While the IBM Watson solution has been effective so far, it remains dependent on a connection to the cloud. And for much of Africa, Internet connectivity is not exactly guaranteed—especially in wildlife areas, where it may be dependent on high-latency satellite connections or sparsely deployed cellular networks. 

Private wireless networks can provide a local backbone for communications, but there still needs to be some sort of back-end connection for a cloud-driven IoT solution to work.

Signal Boost

Alternatively, there are approaches like the Connected Conservation, a program sponsored by Dimension Data and Cisco. Launched in 2015, the program has been using IoT technology to track people near and within the perimeter of a private rhino reserve (its name withheld for security purposes) in South Africa. 

Based on the success of the effort, the program is now preparing to expand to other conservation areas across the continent.

The Connected Conservation program uses fixed cameras, sensors, and a fixed radio network to lock onto and track every human moving through the private reserve. 

Analytics generate alerts based on detected patterns of activity, and rangers are dispatched to intercept intruders.

There is a cloud component, but its role is limited—a local server connects to Microsoft’s Azure services to backup critical data offsite. The whole system is operated instead as a managed service on site for the preserve.

“When the project started, there was very little communication in place as it is a very remote location,” said Wolf Stinnes, solutions architect for Dimension Data. “The only form of communication in place was a person monitoring a laptop with a radio in a centralized room under lock and key.”

Stinnes said that his team worked with Cisco to design and put in place a security solution for the reserve. 

One of the first components, installed in December of 2015, was a new radio-based point-to-point “reserve area network” (RAN). The 50mbps digital radio network, which uses fixed towers along the edge of the reserve, covers the entire perimeter, providing a communications backbone for sharing CCTV video and sensor alerts as well as voice communications.

And besides a spotty connection, the network’s topology had to take into account a number of potential problems. “The biggest challenges are torrential rain, downpours, and lightning, along with the heat,” Stinnes said.

Normal cellular systems couldn’t be used, he noted, because of the low reliability of GSM under those conditions—as well as the ready availability of GSM jamming devices to poachers.

So in addition to the radio network backbone, Cisco and Dimension Data installed wired local area networks at each of the reserve’s four vehicle gates. CCTV cameras, biometric scanners, and other sensors were installed at the gates, as were remote network monitoring systems, routers and switches, and server infrastructure for the managed services. 

The local IT infrastructure allowed the Connected Conservation team to rapidly deploy new testbed equipment and software that will be rolled into future service deployments at new sites. 

And eventually, a wireless network was deployed throughout the reserve to provide secure park-wide Wi-Fi, giving rangers mobile access to data from sensors.

To help boost sensor connectivity throughout the reserve, the reserve’s network is linked to a LoRa wireless wide-area network (LoRaWAN). LoRa is a long range, spread-spectrum wireless technology that uses unlicensed spectrum in the ISM bands (such as 169MHz, 433MHz, 868MHz and 915MHz). 

The LoRaWAN protocol, an open standard, is optimized for low-power devices, so sensors can operate off batteries for years and communicate at ranges of up to 15km. 

While it has a relatively low data rate—27 kbps—that is acceptable for a network handling the relatively low data rate of sensor telemetry (for one example: data from magnetic sensors that were buried in the ground along the perimeter to track vehicle movement).

Other system additions include networked thermal cameras installed along the reserve’s perimeter and fiber-optic acoustic sensors. Each vehicle that enters the reserve is also fitted with a tracking sensor these days. “Using predictive modelling,” said Stinnes, “the analytics team can estimate when an individual or vehicle is expected to exit the reserve.”

Every person who enters the reserve through the gates is tracked. The biometric systems scan the fingerprints of staff, rangers, animal trackers, contractors, and suppliers. 

Visitors have their passports scanned. License plate cameras record the registration of every vehicle that enters, and that info is checked against a national database via a VPN connection, allowing the system to provide analytics on who owns vehicles and the history of their visits. 

All of the gates are linked back to a central control center with 24 video screens over the RAN. “We trained up the people on the ground—the warden, the head of security, and all the control center operators that work for them, to operate the systems themselves,” said Stinnes.

The data gathered from the cameras and sensors are ultimately fed into analytics systems that generate alerts for detected activity that falls outside the norm or possibly indicates a perimeter breach. Drones can be dispatched to the site of alerts for immediate imagery, and an armed ranger team is dispatched by helicopter to intercept poachers.

The entire system has paid off in a big way. 

Since 2015, there has been a reported 96 percent reduction in the number of rhinos poached off the preserve, and no rhinos were taken at all by poachers in 2017. The alert system has decreased the average ranger response time to an incursion from 30 minutes to just seven minutes. 

And because of the elevated response, there has been a 68 percent reduction in the number of incursions into the reserve.

Taking It to Version 2.0

With such initial success, the Connected Conservation program now aims to expand its efforts, working to bring versions of the IoT managed service to parks and reserves in Mozambique, Zambia, and Kenya. 

Each of these locations brings its own challenges, so technology must be adapted to the specific environment and threat for each future site. 

Elephants are the primary concern in Zambia and in Mozambique—where herds are being decimated faster than anywhere else in Africa.

The next Connected Conservation site is an unnamed park in Zambia. The populations in Zambia overall are relatively stable, with a 4.2 percent “carcass ratio” (the measurement of dead elephants to live ones spotted during an aerial survey) out of an elephant count of 21,758 in 2016. 

But things are extremely grim in Sioma Ngwezi National Park on Zambia’s southwest border. There, the survey found an 85 percent carcass ratio, with 17 recently killed elephants for every three found alive. There were also substantial elephant population declines along the lower Zambezi River.

This particular park being targeted for the new Connected Conservation project has to deal with another factor not in play at the initial South African reserve: water. A large lake in the park is regularly fished by people living in the area, but it is also used by poachers to gain quick access to elephants. Given the landscape, a physical perimeter fence is impossible.

To deal with those issues, a network of fixed thermal cameras mounted on radio masts is planned as a virtual fence line for the park’s perimeter to supplement cameras at the park’s entry and exit gates. 

Data from the thermal cameras will be put into analytics to build up a model for identifying patterns of movement in and near the park. The analytics system will generate alerts for the movement of boats across the virtual tripwire and for any nighttime crossings of the park perimeter.

Data and alerts from the system will be sent to a control center set up for a special Zambian marine ranger unit, which will be able to dispatch powerboats to intercept poachers. 

“We’re also working with the Zambian local authorities and the fishing community to create a centralized digital fishing permit system that will monitor individuals who pose as fisherman but are actually poachers,” Dimension Data Group Executive, Bruce Watson said.

Like with the first site, Connected Conservation will also configure outdoor Wi-Fi, mounting access points on the radio masts so that imagery and sensor data can be shared between the security team and rangers and staff on the ground. The Wi-Fi will also provide a secure communications channel for park staff that can’t be easily intercepted by poachers, unlike the handheld radios currently used.

A Scalable Model?

Dimension Data and Cisco have ambitious plans for Connected Conservation beyond rhino and elephant preservation. The project’s ultimate goal is to eliminate all forms of poaching globally, applying IoT devices and other technologies to protect even more vulnerable species around the world. 

Things are increasingly desperate for other endangered animals as well: pangolins in Africa and Asia, tigers, a number of species of sharks, and manta rays are just a few of the endangered and threatened species targeted by poaching, illegal fishing, and animal trafficking. And the Connected Conservation team believes that their IoT service can be used to help protect all of them through cooperation with governments and private organizations.

That’s going to require significant innovation and cooperation, however, since it will also require surveillance of human activity on a vast scale across environments that are difficult at best to surveil. But until the human behaviors that drive demand for these animals are changed, anything to reduce pressure on their populations will help. 

If it takes a few sensors and network entry points, so be it.