Abstract:The design and implementation of a low-cost system for monitoring and remote control of a greenhouse using fuzzy logic is presented. For the control system, an Arduino Mega board was programmed with a fuzzy algorithm to monitor and perform control actions for environmental temperature, soil moisture, relative humidity, and lighting. A website was designed to visualize the main indicators of agricultural interest and to get access to tools such as forced ventilation, misting systems, and sprinkler irrigation. For connectivity to the webpage, an Arduino Ethernet Shield was used. Thus, it was possible to establish a local area network and monitor and control the greenhouse climate variables manually or automatically. The application designed allowed access to the configuration, monitoring, and control of climatic conditions in the greenhouse. The effectiveness of fuzzy logic to control nonlinear systems was therefore verified without the mathematical model of the plant. Thus, the use of resources for a gable roof greenhouse prototype was optimized.

An eight channels subband audio codec is implemented for signals with 44.1 KHz and 16 bits per sample using Matlab. To achieve perfect reconstruction, a two channels QMF filter bank with cutoff frequency = /2 is designed, based on an equiriple filter of 99 order. Seven stages of this bank are used to split the input into eight signals with sample rates from 2.76 to 11 KHz, which are coded from 1 to 16 bits depending on the band energy. To evaluate performance for three tracks in terms of similarity of input and output signals, a Mean Opinion Score (MOS) experiment with fifteen subjects was performed. The Euclidean Distance between spectrums was also measured. Results showed a fair similitude for two tracks and excellent for one. Compression factors above 96% were achieved.Keywords: subband coding, mean opinion score MOS, QMF filter banks, audio compression ResumenSe implementa un codificador subbanda de 8 canales para señales de entrada con frecuencia de muestreo de 44.1KHz y 16 bits por muestra, utilizando Matlab. Para reconstrucción perfecta se diseña un banco de filtros QMF de dos canales con frecuencia de corte = /2, a partir de un pasobajo equiriple de orden 99. Se utilizan siete etapas del banco diseñado para generar ocho señales con frecuencias de muestreo de 2.76 a 11 KHz, las cuales son codificadas con tamaños de palabra de 1 a 16 bits, dependiendo de la energía de la banda. Para hacer una evaluación subjetiva del desempeño, se mide la similitud de la señal original con la codificada para tres pistas musicales, realizando una prueba MOS a quince sujetos. Como medición objetiva se calcula distancia euclidiana entre espectros de la señal original y la reconstruida. Los resultados mostraron similitud aceptable para dos pistas y excelente para una. Se logran factores de compresión superiores al 96%.