LoRaWAN Scalability Under Worst-Case ALOHA: A Single-Channel Stress Test for Dense IoT Networks
This study isolates single-channel LoRaWAN performance under worst-case pure ALOHA conditions to identify a critical scalability threshold of 50–60 nodes where packet delivery sharply degrades, revealing that confirmed transmission modes offer initial reliability gains but ultimately incur prohibitive energy costs and reduced performance at high densities, while node mobility proves negligible in collision-dominated regimes.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
In the quiet corners of the internet, where sensors monitor soil moisture, track livestock, or report air quality, a vast network of tiny devices communicates without wires. These devices rely on a technology called LoRaWAN, designed to send small bursts of data over long distances while using very little battery power. Imagine a room full of people trying to talk to a single listener at the front. If everyone waits their turn, the conversation flows smoothly. But if everyone speaks at once, hoping their voice cuts through the noise, the result is a chaotic jumble where no one is understood. This is the fundamental challenge of the network: how many devices can speak before the airwaves become too crowded to hear anything? For years, engineers have known that these networks work well for sparse, slow-moving data, but as applications grow more demanding—requiring the transmission of images or rapid updates from moving vehicles—the limits of this system have remained unclear.
A team of researchers from Universiti Malaysia Sarawak decided to find the breaking point of this system under the most difficult conditions imaginable. They did not look at a network with many gateways or multiple frequency channels, which would spread the traffic out and hide the true limits. Instead, they isolated a single gateway listening on a single channel, forcing hundreds of devices to compete for the same tiny slice of radio space. They simulated a scenario where devices sent data continuously, mimicking the stress of high-throughput applications like streaming images or tracking fast-moving objects. By pushing the network to its absolute limit, they sought to discover exactly when the system would begin to fail and what that failure looked like.
The researchers built a detailed computer simulation to test this stress, running over 1.6 million individual packet transmissions across sixty different scenarios. They varied the number of devices from just one up to two hundred, and they moved these devices at speeds ranging from a standstill to seventy-two kilometers per hour. They also tested two different ways of sending data: one where the device sends a message and waits for a "got it" signal before sending the next, and another where it sends the message once and moves on without waiting. The goal was to see how many devices could be supported before the success rate of these messages dropped below an acceptable level.
The results revealed a sharp and sudden tipping point. When the network had fifty devices or fewer, it performed almost perfectly, delivering nearly all messages successfully regardless of how fast the devices were moving. However, as the number of devices climbed from fifty to sixty, the performance collapsed. The success rate of message delivery plummeted from nearly ninety-eight percent down to roughly seventy-eight percent in that narrow window. This transition was not a slow decline but a steep drop, indicating that the network had hit a hard ceiling. Beyond sixty devices, the situation worsened rapidly, with the success rate falling to less than thirty-four percent at one hundred devices and down to twelve percent at two hundred.
Surprisingly, the speed of the moving devices made almost no difference to this outcome. Whether the devices were stationary or racing along at highway speeds, the network failed at the same density. The researchers found that once the number of devices became high enough to cause frequent collisions—where two messages arrive at the same time and cancel each other out—the movement of the devices became irrelevant. The chaos of the crowd overwhelmed any benefits or problems caused by motion. In this crowded environment, the primary cause of failure was not a weak signal or a device moving out of range, but simply too many voices trying to speak at once.
The study also uncovered a critical flaw in how the network handles reliability when it is already struggling. In the "confirmed" mode, where devices wait for an acknowledgment, the system initially seemed to offer a safety net. At fifty devices, this mode improved reliability significantly compared to the "unconfirmed" mode. However, as the network became more crowded, this safety net turned into a trap. The devices, failing to get their messages through, would keep trying to resend them, creating even more traffic and more collisions. By the time the network reached two hundred devices, the confirmed mode was actually performing worse than the simpler unconfirmed mode, delivering fewer messages while consuming nearly eight times more energy. The very mechanism designed to ensure reliability had become the cause of its own destruction.
These findings provide a clear, conservative benchmark for network designers. The research suggests that for a single-channel gateway operating under these demanding conditions, the safe limit is around fifty devices. Beyond this point, the network enters a zone of rapid degradation where adding more devices yields diminishing returns and eventually negative results. The study proposes a way to think about larger networks by separating the fundamental limit of a single channel from the benefits of adding more hardware or using smarter protocols. While multi-channel gateways can support many more devices by spreading the load, the underlying limit of the single channel remains a fixed constraint. For anyone planning to deploy dense networks of sensors, especially those that need to send large files or track movement, this research offers a vital warning: there is a hard limit to how much chaos a single radio channel can handle, and pushing past it breaks the system faster than anyone expected.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.