Without the movement of arms and legs, the patient suffering from a stroke (Cerebral Vascular Accident) controls, through brain impulses, his own wheelchair, with a simple glance in the direction he wants to go. In another situation, a person can communicate via cell phone, including accessing the internet, anywhere on the planet, even in the most remote places where signals are very weak. In another scenario, clear and detailed images inform geology professionals about the possible existence of structures favorable to the presence of gas and oil. “The potential for applying the signal processing technique is enormous”, guarantees Unicamp professor João Marcos Travassos Romano when citing examples of the main projects developed in the laboratory he coordinates.

The professor at the Faculty of Electrical and Computer Engineering (FEEC), together with a team of undergraduates, master's students, doctoral students and post-doctoral students - including professors from the same unit Renato da Rocha Lopes and Romis Ribeiro de Faissol Attux - cooperates with several groups of researchers from the University and other institutions. The objective is to formulate signal processing techniques for the development of projects with an impact on society.

“Here at DSPCom [Signal Processing Laboratory for Communications] we work in the areas of adaptive and unsupervised systems, 'machine learning' and communications theory. The main applications of signal processing are in modern telecommunications systems, seismic processing for gas and oil exploration and the processing of audio, image and biological signals, as in the case of human-machine communication”, specifies Romano .

The teacher explains that signal processing or Signal processing, from the English term, is based on different fundamental areas of science, such as physics, mathematics, and computing. The technique extends to applications as diverse as medicine and geology, he adds.

Romano explains that signals are basically mathematical functions that, in the population's daily lives, can represent the most varied quantities, indicating, for example, audio, image, distance, temperature, pressure, etc. O Signal processing seeks to extract information from these signals – through mathematical and physical calculations – to make them more suitable for varied applications.  

“It is a fundamental area for training electrical engineers. It uses tools from classical and applied mathematics; engineering; physics and computing. It is also multidisciplinary in its applications. The examples are in our laboratory: we collaborate with groups from Unicamp and other universities, including partnerships with industry institutions, including Petrobras, Fundação CPqD, Ericsson and Transtel”, he explains.

Artificial control

Researcher Thiago Bulhões da Silva Costa - guided in his ongoing master's degree by professor Romis Attux - has been investigating and carrying out experiments to extract information from brain signals to assist in the development of technologies for the locomotion of patients with loss of movement. “We try to identify brain patterns, via signal processing, so that the individual can, for example, give commands in a wheelchair, just by imagining the movement”, explains the scholar.

Your master's degree is part of a broad interdisciplinary project called Destine (Development of Information Technologies for Neurology). The initiative, coordinated by professor José Raimundo de Oliveira, from FEEC, involves a series of professors and researchers from Unicamp and other institutions. Romis Attux clarifies that Destine also extends to the area of ​​robotics, with the development of 'smart' sensors to help people with a wide range of everyday tasks.

“We transform a raw brain signal into a control signal, carrying out a series of processing. This control signal, in turn, will be the input to a robotic system. The idea is to have a brain-computer interface [BCI for Brain Computer Interface] working at acquisition time”, says Attux.

In addition to the DSPCom researchers, the Destine project involves the participation of Eleri Cardozo, Fernando José Von Zuben and Ricardo Ribeiro Gudwin, (all from FEEC); Leonardo Tomazeli Duarte (Faculty of Applied Sciences of Limeira); Roberto José Maria Covolan and Gabriela Castellano (Institute of Physics 'Gleb Wataghin'); Li Li Min and Cynthia Resende Campos Herrera (Faculty of Medical Sciences); Paula Teixeira Fernandes Boaventura (Faculty of Physical Education); Ricardo Suyama (Federal University of ABC) and Eliane Gomes Guimarães (Renato Archer Information Technology Center). The project receives resources from the Financier of Studies and Projects (Finep), a Brazilian funding agency linked to the Ministry of Science, Technology and Innovation (MCTI).

The best image

Using signal processing techniques, DSPCom researchers work to provide images, richer in details and without 'noise', to Petrobras technicians. Such images may indicate geological structures capable of accumulating oil and gas. The objective is to guide the Brazilian oil company in drilling a certain area, a costly and complex activity.

“The partnership began in 2011, when Cepetro [Center for Petroleum Studies at Unicamp] organized a workshop with the presence of Petrobras professionals. Two electrical professors work on our project team, Renato Lopes and I; a professor at FCA, Leonardo Duarte; three postdocs; and five doctoral students, one of them from France, from the École de Mines in Paris. It is a broad agreement, coordinated at Unicamp by professor Martin Tygel, from Imecc [Institute of Mathematics, Statistics and Scientific Computing]”, informs João Marcos Romano.

“Through signal processing algorithms, we seek to generate images of seismic data with the highest possible resolution. These images will support oil technicians in the task of identifying reservoirs”, adds professor Renato Lopes.

The best signal

Guided by professor Renato Lopes, researcher Marco Aurélio Cazarotto Gomes has been studying, in his doctorate, methods to promote a type of 'interference cancellation' in communication systems. The focus is to make communication systems suitable for transmitting data at speeds close to their full capacity.

“This can be used in any system capable of transmitting digital information from point A to point B. The intention is to find out how much the signal transmission rate can be improved. Cellular systems still have a lot to gain until they approach theoretical limits for transmission speed, for example. In general, communication systems, as they currently operate, are a relatively long way from reaching this limit due to complexities and a series of technical problems. Our objective is to propose techniques that overcome these barriers”, says the advisor.

Books

Three relevant works in the field of signal processing have already been generated by research carried out at the Laboratory, highlights Romano. The most recent, published last year, is Chaotic Signals in Digital Communications (Chaotic signals in digital communications, translated into Portuguese), organized by professor Romis Attux in partnership with Marcio Eisencraft, from the Polytechnic School of the University of São Paulo (USP); and Ricardo Suyama, from the Center for Engineering, Modeling and Applied Social Sciences at the Federal University of ABC.

The most emblematic work of the group's activities is Unsupervised Signal Processing: Channel Equalization and Source Separation (Unsupervised signal processing: channel equalization and source separation), 2011. The book is authored by João Marcos Romano, Romis Attux, Ricardo Suyana and Charles Casimiro Cavalcante, from the Federal University of Ceará (UFC), the latter ex -DSPCom researchers.

The other title is Independent Component Analysis and Signal Separation (Independent analysis of components and signal separation), published in 2009. The work is organized by João Marcos Romano; Tulay Adali, University of Maryland Baltimore County; Christian Jutten, from Université Joseph Fourier; and Allan Kardec Duailibe Barros, from the Federal University of Maranhão (UFMA).