Connectivity availability and costs
For years the main limitation to scale up in extensive agriculture projects consisted of the availability of connectivity and its costs. Connectivity of $2 or $3 per month for a sensor does not seem to be an obstacle. However, it is not until it is scaled to 1,000, 10,000, or 100,000 sensors that the real cost of a large-scale expansion of this technology is understood.
During the last five years the availability of radiofrequency protocols known as LPWAN (Low Power Wide Area Networks) such as SigFox or LORA and networks such as NB-IoT and LTE-M built on traditional cellular bands, managed to provide feasible coverage and reasonable costs to scaled-up projects of thousands of hectares. Standard connectivity in these services of $1 to $2 per year ($0.08 - $0.16 per month) if they allow a greater ubiquity of the technology in the agricultural sector.
A great challenge when implementing connectivity networks for large extensions of land is that the initial investment in telecommunications equipment can tend to be prohibitive to justify a return on the project. That is why the best option for most cases is to look for public network operators who finance this investment and distribute its cost through the subscription of all available clients. In this sense, SigFox, LORAWan, and telephone operators (with bands specifically for IoT) once again become the best technical and financial option.
The full cost of sensors and TCO
Proof-of-concept implementations that have been successful from a conceptual point of view, but not as a scale project, are abundant in the market. Usually, this happens because the selection of hardware sensors is made with little electronic technical knowledge and often with no business background, thus making the individual sensor cost-prohibitive to scale any project to thousands of square kilometers of land.
The sensor itself carries several technical layers that can make it more or less cost-efficient. For example, a simple humidity transducer (transducer: device that translates a physical variable into an electrical one) can cost as little as $10 individually. However, shod with the wrong technology, you may have to put up with an expensive wireless transmitter, a bank of special batteries, antennas, and even a solar panel; turning what could have been a scalable system into a $400+ sensor demo.
In sensors design, it is therefore important to clearly understand the characteristics of the use of the sensor in the field and to choose the most limited and balanced transmission and energy mechanisms for the business value of the data to be measured.
The vast majority of agricultural applications fall into the IoT monitoring layer and only a few require control mechanisms. An example of this is self-driving systems for harvesting equipment or intelligent irrigation systems. To a greater or lesser degree, when we need to control remote equipment, response time can be critical in its operation. In immediate response systems, such as alerts, or high-precision systems such as self-guided GPS equipment, network latency, and synchrony become critical in the implementation of an IoT solution. This means that expansion often depends on the availability and cost of high-speed communication networks on the farm. In this case, LTE and soon 5G are the technologies called to solve this problem.
Finally, a non-technical obstacle to scaling agricultural IoT projects is vandalism. In this type of project, infrastructure is placed outdoors and prone to vandalism. That is why it is increasingly important for manufacturers to incorporate guard mechanisms that emit anti-vandalism alarms in the event of disassembly or improper manipulation.
Maintenance and updating
Another similar big problem is the regular maintenance of an infrastructure spread over hundreds of hectares. The ability to remotely monitor and maintain these devices is something that should be natively incorporated into all elements of the architecture. Ideally, the sensor and gateways should be designed to be able to monitor and update remotely from the platform without the need to go to the field one by one.