Journal of Ubiquitous Systems & Pervasive Networks Volume 11, No. 2 (2019) pp. 01-08 * Corresponding author. Tel.: +34973702715 Fax: +34973702700; E-mail: mtresanchez@diei.udl.cat © 2019 International Association for Sharing Knowledge and Sustainability. DOI: 10.5383/JUSPN.11.02.001 1 An inexpensive wireless smart camera system for IoT applications based on an ARM Cortex-M7 microcontroller M. Tresanchez*, A. Pujol, T. Pallejà, D. Martínez, E. Clotet, J. Palacín Department of Computer Science and Industrial Engineering, University of Lleida, Lleida, Spain, 25001 Abstract This paper presents a new proposal of a low-cost and low-power wireless smart camera device, designed to be used as an inexpensive smart image acquisition edge node for Internet of Things (IoT). Taking the advantages of the ARM Cortex- M7 microcontrollers, a high-performance STM32F7 microcontroller is used as a main digital signal processor (DSP) to acquire and transfer images at full rate, keeping the core idle to do real-time image processing. In order to design a very compact, low-cost and low-power wireless image sensor node, the DSP is supported by an embedded 1.3 Mega Pixel CMOS camera sensor and a low consumption 2.4GHz Wi-Fi module. This setup allows a custom build-in image processing algorithm for specific IoT sensing applications or a simply setup as low-cost streaming Motion JPEG (MJPEG) wireless camera node. In this work, both configurations have been tested and analyzed. Results show that the device can acquire and process images simultaneously at a full rate (30 fps) and the MJPEG transmission reaches 7.23 fps with a QVGA resolution. Keywords: IoT node, Image sensor, Video streaming, ARM, Cortex-M7 1. Introduction Image sensors are an important type of sensor in several emerging Internet of Things (IoT) applications. They provide detailed environment information using complex technology yet having a highly competitive market price due to their standardization. Their basic application as a camera system for monitoring or supervising a location through video streaming can be extended when they are combined with image processing techniques [1-4]. In IoT, image sensors may be applied in many fields, some recent contributions use image sensors for surveillance video systems in smart cities [5], for personalized healthcare systems, for recognizing and understanding human activities [6, 7] or for video streaming systems in vehicular IoT [8]. Image acquisition systems require a high performance computation in order to handle the large amount of data. Furthermore, if they have to process and transmit image information through a wide network, more powerful and expensive Digital Signal Processors (DSP) are needed and, consequently, the current consumption increases. This is an important drawback in IoT wireless sensors network management [9] where it is crucial to have low-cost and low- power nodes because usually there are many and powered on all time. Additionally, another relevant factor to consider is the node size, weight and the hardware complexity for final product manufacturing. In [9] and [10] a more complex hardware is used based on System-On-Chip (SoC) which needs and operating system, making it less robust and requires a delicate assembly process that hinders its integration in a final IoT sensor product. In this work, an embedded image acquisition system based on the new high-performance Cortex-M7 microcontroller and a compact CMOS camera module is proposed as a smart wireless IoT image sensor node. This proposed device has a very low power consumption, ideal in IoT environments, and requires very few electronic components which help to achieve a small size product at a competitive production cost. In order to balance the performance and consumption, depending on the application level requirements, the selected hardware allows to configure different low-power modes. This is essential in outdoor IoT applications where the device uses external energy resources like solar energy [11]. In this work, the proposed system takes advantage of the microcontroller internal hardware peripherals to acquire and transfer images at the maximum wireless adapter throughput with a minimum computational load. This means that no software or operating system is required for the video