RTOS vs Linux: The IoT battle extends from software to hardware

There’s certainly something happening regarding operating systems in the Internet of Things (IoT). Chips are getting smarter, devices are getting smaller, and speeds are getting faster. As a result, device developers are more often experimenting with their operating system of choice, moving away from Linux and toward real-time operating systems (RTOS).

This is an evolution on two fronts. On the software side, applications requiring low latency and deterministic responses are turning to Zephyr, FreeRTOS, and ThreadX. And now, on the hardware side, we’re seeing more chip manufacturers entering the fray with RTOS-specific hardware that rivals or surpasses performance of entry-level Linux boards. This is a big deal since these chips optimize hardware-software integration, creating a more thorough ecosystem for purpose-built solutions with RTOS.

This is an ongoing operating system battle and developers must proceed cautiously. On the one hand, RTOS promises real-time performance, attack surface reduction, and newfound efficiency. On the other, switching operating systems demands a close eye on device stability and development complexity. Let’s explore.

What’s happening in device software, hardware, and operating systems

As I wrote a few months back, Linux is an excellent operating system but it’s not always ideal for today’s smaller and simpler devices. Many “things” like video doorbells don’t require a full-fledged OS, and Linux’s growing complexity can make it vulnerable to undetected malicious code. Meanwhile, RTOS has rapidly evolved, offering compelling alternatives for IoT applications.

These real-time systems are specifically designed to handle tasks with precision and ensure that critical processes execute on time. Unlike traditional operating systems, which can be bulky and complex, RTOS typically occupies just a few megabytes of memory, making it an ideal fit for smaller devices and critical applications.

Recognizing this shift, hardware manufacturers are co-signing the move toward RTOS. Heavy hitters like Microchip Technology and NXP Semiconductors offer a variety of microcontrollers compatible with RTOS. Perhaps the best example is from Espressif and its ESP32 series. This system-on-chip is optimized for RTOS and features a robust core, decent memory, and wireless connectivity. While it may not reach gigahertz speeds, operating at 400-600 MHz allows these chips to compete effectively with low-end Linux solutions, paving the way for more efficient and responsive IoT applications.

Why this evolution matters

By supporting RTOS-optimized chips, these manufacturers are providing purpose-built hardware that enhances performance and reduces latency in critical applications. As a result, this empowers developers to more easily transition to RTOS, potentially reshaping industries reliant on real-time processing.

This transformation is particularly evident when comparing RTOS to Linux. Unlike Linux, which maintains a separation between user space and kernel space, RTOS applications can communicate directly with hardware. This eliminates the overhead of context switches and data copying, resulting in faster response times. The reduced processing layers in RTOS minimize the physical and logical distance between application code and hardware resources. This is something I've witnessed firsthand in a video transfer project – Linux’s context switching and memory copying between user and kernel space caused performance issues that could have been avoided entirely with RTOS.

Further, this streamlined architecture allows for greater efficiency in resource utilization. By minimizing overhead, RTOS optimizes performance despite limited processing power and memory, an important development across devices operating under strict limitations in robotics, industrial automation, and real-time control systems.

RTOS vs Linux: Understanding the key trade-offs

Of course, RTOS is no silver bullet. There are trade-offs in making the switch and developers need to be aware of what it can mean for their devices. For example, while it offers better performance for specific applications, RTOS lacks the isolation between applications that Linux provides, which can be a downside in terms of system stability.

Developers must weigh factors like ecosystem support, development complexity, and scalability when choosing between Linux and RTOS. The former has a vast ecosystem of libraries, tools, and third-party support, and ecosystems for the latter are playing catch up in this regard. If opting for RTOS, remember that they're not all created equal -- consider modularity, connectivity, and GUI capabilities when choosing the best RTOS for your device or project.

Interestingly, some manufacturers are straddling both sides of the operating system spectrum. While many chips are optimized for RTOS or Linux, companies like ST Microelectronics and Texas Instruments are developing versatile chips that can accommodate multiple operating systems. In theory, this lets developers use RTOS for real-time tasks and Linux for complex, non-time-critical operations. In practice, resource sharing requires significant development skills and should only be attempted by teams with deep technical experience.

Whatever happens next and whatever you decide, this is an exciting time in device development. Diverse use cases deserve diverse operating systems and it’s heartening to see hardware producers back this software evolution. Embracing this shift will empower industries to innovate and adapt, leading to smarter and more efficient devices for everyone.

Image credit: Jirsak / Shutterstock

Carsten Rhod Gregersen is the CEO and founder of Nabto, a company that provides a P2P-based platform for IoT devices