Design and Implementation of an LSTM Model with Embeddings on MCUs for Prediction of Meteorological Variables

The use of recurrent neural networks has proven effective in time series prediction tasks such as weather. However, their use in resource-limited systems such as MCUs presents difficulties in terms of both size and stability with longer prediction windows. In this context, we propose a variant of th...

Full description

Saved in:
Bibliographic Details
Main Authors: Jhan Piero Paulo Merma Yucra, David Juan Cerezo Quina, German Alberto Echaiz Espinoza, Manuel Alejandro Valderrama Solis, Daniel Domingo Yanyachi Aco Cardenas, Andrés Ortiz Salazar
Format: Article
Language:English
Published: MDPI AG 2025-06-01
Series:Sensors
Subjects:
Online Access:https://www.mdpi.com/1424-8220/25/12/3601
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849425998850818048
author Jhan Piero Paulo Merma Yucra
David Juan Cerezo Quina
German Alberto Echaiz Espinoza
Manuel Alejandro Valderrama Solis
Daniel Domingo Yanyachi Aco Cardenas
Andrés Ortiz Salazar
author_facet Jhan Piero Paulo Merma Yucra
David Juan Cerezo Quina
German Alberto Echaiz Espinoza
Manuel Alejandro Valderrama Solis
Daniel Domingo Yanyachi Aco Cardenas
Andrés Ortiz Salazar
author_sort Jhan Piero Paulo Merma Yucra
collection DOAJ
description The use of recurrent neural networks has proven effective in time series prediction tasks such as weather. However, their use in resource-limited systems such as MCUs presents difficulties in terms of both size and stability with longer prediction windows. In this context, we propose a variant of the LSTM model, which we call SE-LSTM (Single Embedding LSTM), which uses embedding techniques to vectorially represent seasonality and latent patterns through variables such as temperature and humidity. The proposal is systematically compared in two parts: The first compares it against other reference architectures such as CNN-LSTM, TCN, LMU, and TPA-LSTM. The second stage, which includes implementation, compares it against the CNN-LSTM, LSTM, and TCN networks. Metrics such as the MAE and MSE are used along with the network weight, a crucial aspect for MCUs such as an ESP32 or Raspberry Pi Pico. An analysis of the memory usage, energy consumption, and generalization across different regions is also included. The results show that the use of embedding optimizes the network space without sacrificing the performance, which is crucial for edge computing. This research is part of a larger project, which focuses on improving agricultural monitoring systems.
format Article
id doaj-art-181734161a4948eaba66eeb9fece1fbe
institution Kabale University
issn 1424-8220
language English
publishDate 2025-06-01
publisher MDPI AG
record_format Article
series Sensors
spelling doaj-art-181734161a4948eaba66eeb9fece1fbe2025-08-20T03:29:35ZengMDPI AGSensors1424-82202025-06-012512360110.3390/s25123601Design and Implementation of an LSTM Model with Embeddings on MCUs for Prediction of Meteorological VariablesJhan Piero Paulo Merma Yucra0David Juan Cerezo Quina1German Alberto Echaiz Espinoza2Manuel Alejandro Valderrama Solis3Daniel Domingo Yanyachi Aco Cardenas4Andrés Ortiz Salazar5Professional School of Electronic Engineering, Universidad Nacional de San Agustín de Arequipa, Arequipa 04002, PeruProfessional School of Electronic Engineering, Universidad Nacional de San Agustín de Arequipa, Arequipa 04002, PeruProfessional School of Electronic Engineering, Universidad Nacional de San Agustín de Arequipa, Arequipa 04002, PeruProfessional School of Telecommunications Engineering, Universidad Nacional de San Agustín de Arequipa, Arequipa 04002, PeruProfessional School of Electronic Engineering, Universidad Nacional de San Agustín de Arequipa, Arequipa 04002, PeruDepartment of Computer Engineering and Automation, Federal University of Rio Grande do Norte (DCA-UFRN), Natal 59072-970, RN, BrazilThe use of recurrent neural networks has proven effective in time series prediction tasks such as weather. However, their use in resource-limited systems such as MCUs presents difficulties in terms of both size and stability with longer prediction windows. In this context, we propose a variant of the LSTM model, which we call SE-LSTM (Single Embedding LSTM), which uses embedding techniques to vectorially represent seasonality and latent patterns through variables such as temperature and humidity. The proposal is systematically compared in two parts: The first compares it against other reference architectures such as CNN-LSTM, TCN, LMU, and TPA-LSTM. The second stage, which includes implementation, compares it against the CNN-LSTM, LSTM, and TCN networks. Metrics such as the MAE and MSE are used along with the network weight, a crucial aspect for MCUs such as an ESP32 or Raspberry Pi Pico. An analysis of the memory usage, energy consumption, and generalization across different regions is also included. The results show that the use of embedding optimizes the network space without sacrificing the performance, which is crucial for edge computing. This research is part of a larger project, which focuses on improving agricultural monitoring systems.https://www.mdpi.com/1424-8220/25/12/3601edge computingembeddingsLSTMclimate prediction
spellingShingle Jhan Piero Paulo Merma Yucra
David Juan Cerezo Quina
German Alberto Echaiz Espinoza
Manuel Alejandro Valderrama Solis
Daniel Domingo Yanyachi Aco Cardenas
Andrés Ortiz Salazar
Design and Implementation of an LSTM Model with Embeddings on MCUs for Prediction of Meteorological Variables
Sensors
edge computing
embeddings
LSTM
climate prediction
title Design and Implementation of an LSTM Model with Embeddings on MCUs for Prediction of Meteorological Variables
title_full Design and Implementation of an LSTM Model with Embeddings on MCUs for Prediction of Meteorological Variables
title_fullStr Design and Implementation of an LSTM Model with Embeddings on MCUs for Prediction of Meteorological Variables
title_full_unstemmed Design and Implementation of an LSTM Model with Embeddings on MCUs for Prediction of Meteorological Variables
title_short Design and Implementation of an LSTM Model with Embeddings on MCUs for Prediction of Meteorological Variables
title_sort design and implementation of an lstm model with embeddings on mcus for prediction of meteorological variables
topic edge computing
embeddings
LSTM
climate prediction
url https://www.mdpi.com/1424-8220/25/12/3601
work_keys_str_mv AT jhanpieropaulomermayucra designandimplementationofanlstmmodelwithembeddingsonmcusforpredictionofmeteorologicalvariables
AT davidjuancerezoquina designandimplementationofanlstmmodelwithembeddingsonmcusforpredictionofmeteorologicalvariables
AT germanalbertoechaizespinoza designandimplementationofanlstmmodelwithembeddingsonmcusforpredictionofmeteorologicalvariables
AT manuelalejandrovalderramasolis designandimplementationofanlstmmodelwithembeddingsonmcusforpredictionofmeteorologicalvariables
AT danieldomingoyanyachiacocardenas designandimplementationofanlstmmodelwithembeddingsonmcusforpredictionofmeteorologicalvariables
AT andresortizsalazar designandimplementationofanlstmmodelwithembeddingsonmcusforpredictionofmeteorologicalvariables