Harvest-Aware Real-Time Scheduling for Solar-Powered Edge Nodes under African Climatic Conditions

In the vast, sun-drenched landscapes of sub-Saharan Africa, millions of people live without access to a reliable electrical grid. To bring power to these remote communities, engineers are increasingly relying on decentralized solar systems: small panels that charge batteries to run everything from water pumps to village mini-grids. But these systems need a brain to manage them, a tiny computer that monitors the flow of electricity and reports faults. The challenge is that this computer must run on the same solar power it is monitoring, often in places where the sun is unpredictable. It faces a daily struggle: the sun rises and sets, clouds roll in for days at a time, and seasons shift from dry heat to heavy rain. If the computer runs out of energy, the system goes dark, and critical warnings go unheard. The core problem for these devices is not just having enough power on average, but knowing exactly how much energy to save for a rainy day when the sun refuses to shine.

A researcher at the University of Lagos has tackled this problem by developing a new way to schedule tasks on these solar-powered computers. They focused on a specific type of device that runs two kinds of jobs: critical tasks that must happen on time, like checking for faults or sending alarm signals, and optional tasks that are useful but can be skipped if energy is tight, like uploading large data files or running complex analytics. The researcher created a rule-based system that acts as a strict gatekeeper. Before the computer starts an optional job, it checks its battery level against a pre-calculated safety margin, which they call a "reserve." If the battery does not have enough extra energy to cover this reserve after the job is done, the optional job is rejected immediately. This ensures that the critical tasks always have enough power to finish, even if the sun disappears for several days, provided the battery is large enough to hold the calculated reserve.

The researcher proved mathematically that if this reserve is set high enough to cover the worst possible energy shortage the device might face, and if the battery capacity is sufficient to store that reserve, the critical tasks will never fail. They showed that this reserve can be calculated precisely using decades of weather data, without needing to predict the weather in real-time. To test their idea, they gathered twenty-five years of satellite data on sunlight, temperature, and rain from six different locations across Africa, ranging from the humid coast of Nigeria to the arid desert of Egypt and the Mediterranean climate of South Africa. They simulated how a typical monitoring node would behave over these twenty-five years under different scheduling rules. The results were striking: the amount of energy needed to guarantee safety varied wildly depending on the location. For the same device design, the required safety reserve in Cape Town was 3.6 times larger than in the desert city of Aswan, simply because of the different patterns of cloud cover and day length in those regions.

The study also revealed that relying on short-term data to set these safety margins is dangerous. When the researcher tried to calculate the reserve using only one year of data, that estimate was too low for nearly half of the subsequent years. In other words, a device sized based on a single "typical" year would likely run out of power during the next few years. The most effective strategy they found was to use a "seasonal reserve," which adjusts the safety margin based on the time of year. For example, in Cape Town, where winter storms can last for days, the system automatically holds back more energy during those months, allowing it to serve more optional tasks during the sunny summer months. This approach allowed the device to complete almost all of its optional work while still protecting the critical tasks, a feat that other common scheduling methods failed to achieve.

One of the most dramatic findings came from looking at the worst-case scenario in the entire dataset: a period in June 2008 in Cape Town where the sun was hidden for nearly four days. During this event, standard scheduling methods that tried to use every bit of energy failed completely, causing the critical tasks to miss their deadlines. The new reserve-based system, however, held firm, provided the battery was large enough to store the required reserve. Even though it had to reject many optional jobs to survive the storm, it kept the critical monitoring functions running without interruption. The researcher found that while the reserve-based method sometimes had to say "no" to optional work more often than other methods, it was the only approach that never let the critical system fail, as long as the storage battery was large enough to hold the calculated reserve.

The study also looked at how dust and dirt on the solar panels affect the system. In dry regions like Aswan, where rain is rare to wash the panels clean, dust can block a significant amount of sunlight. The researcher found that if the safety reserve is calculated without accounting for this dust, the system can fail even in sunny deserts. They concluded that for these devices to work reliably, the safety margin must be tailored to the specific local climate, the specific history of weather at that site, and even the local conditions like dust accumulation. They demonstrated that with a battery capacity of just one watt-hour, a device using their seasonal reserve policy could operate for twenty-five years without missing a single critical deadline at five of the six sites they studied. At the most challenging site, Cape Town, it missed only four days, all of which occurred during an extreme weather event that was far worse than anything seen in the previous twenty-four years.

This work moves the field beyond simple averages. It shows that for solar-powered devices in Africa, the key to reliability is not just having a big battery, but having a smart way to decide what to do with the energy inside it. By using a simple rule that reserves a portion of the battery for the worst possible day, these small computers can keep working when the sun goes away, provided the battery is large enough to hold that reserve. The researcher made their tools and data available to the public, allowing engineers to calculate the exact safety margin needed for any location on the continent. Their findings suggest that the future of reliable off-grid monitoring lies in respecting the local climate's history, using long records to plan for the worst, and accepting that sometimes, to keep the lights on, you must be willing to turn off the extras.