Embedded systems—such as Internet of Things (IoT) devices and single-board computers—often operate under strict memory and processing limitations, making efficient resource management essential. Linux, with its flexibility and cost-effectiveness, has become a popular choice for these platforms. Variants like Ubuntu Core, Raspberry Pi OS, BalenaOS, and OpenWrt are widely deployed across embedded devices. However, Linux’s default memory allocator, ptmalloc, frequently struggles to meet diverse application demands. Alternative allocators such as jemalloc, tcmalloc, and mimalloc offer improvements, but their large memory footprints, complex designs, and eventual performance degradation limit their suitability for lightweight environments. This gap underscores the need for more efficient, compact solutions.
To address this challenge, researchers led by Dr. Hwajung Kim, Assistant Professor of Smart ICT Convergence Engineering at Seoul National University of Science and Technology (SeoulTech), Republic of Korea, developed LWMalloc—a lightweight, high-performance dynamic memory allocator tailored for resource-constrained systems. Their findings were published in the IEEE Internet of Things Journal on 15 June 2025.
LWMalloc employs a streamlined data structure, a deferred coalescing (DC) policy, and dedicated small chunk pools to enhance allocation efficiency. The compact data structure reduces metadata overhead, while the DC policy defers unnecessary operations until allocation, lowering execution overhead and maintaining low response times. Meanwhile, fixed-size small chunk pools handle frequent small requests in constant time (O(1)), optimizing allocation patterns common in embedded workloads.
Extensive testing on Raspberry Pi Zero W, Raspberry Pi 4, and Jetson Orin Nano demonstrated LWMalloc’s advantages. “Our proposal outperforms ptmalloc, achieving up to 53% faster execution time and 23% lower memory usage. With just 530 lines of code and a 20 KB footprint—compared to ptmalloc’s 4,838 lines and 116 KB—LWMalloc offers a compact yet powerful solution,” explains Dr. Kim.
The allocator is particularly valuable for IoT and embedded systems with stringent constraints, including smart TVs, set-top boxes, home appliances, mobile and wearable devices, automotive systems, and edge computing nodes running AI or data-intensive tasks.
Beyond performance gains, efficient allocators like LWMalloc promise broader impacts: longer device lifespans, reduced energy consumption, and support for more advanced applications on low-power hardware. “This could make high-performance features accessible on affordable consumer devices, cut down on e-waste, and boost the reliability of everyday embedded systems. As IoT and edge computing scale, such lightweight allocators will be critical for ensuring sustainability and responsiveness worldwide,” concludes Dr. Kim.
Author: shadjavist
Humble gentleman who never believes in impossibility
