Background. Internet of Things (IoT) devices are more prevalent than ever, handling all sorts of sensitive data. This gives rise to the need for encryption algorithms designed to operate in such resource-constrained environments, where processing power, memory, and energy are minimal. This need led the National Institute of Standards and Technology (NIST) to begin the process of standardizing a new encryption algorithm intended for such devices. After several rounds, 10 algorithms remained as the finalists.

Objectives. The objective of this thesis is to evaluate and analyze the performance of these finalist algorithms based on three metrics: throughput, memory utilization, and power consumption. 

Methods. To achieve this objective, three experiments were conducted. Wherein the encryption and decryption speed, stack and heap utilization, and power consumption were measured when the algorithms were executed on an Arduino UNO R4 WiFi, an Arduino NANO ESP32, and an Arduino GIGA R1 WiFi.

Results. In the throughput experiment, the SPARKLE algorithm outperformed the rest of the algorithms. TinyJambu was the algorithm that achieved the overall lowest memory footprint. And in terms of power consumption, Xoodyak was the algorithm that consumed the least power.

Conclusions. The results of the throughput and memory experiments were clear and directly addressed the objectives of this study regarding these aforementioned metrics. The results of the power consumption experiment were inconclusive, with minimal differences between the algorithms with low power consumption. The Power Profiler Kit II, used to measure power consumption, has a ± 20\% accuracy, which further limits the ability to draw accurate conclusions. However, this thesis was able to identify the Elephant algorithm as the one with the highest power consumption.