ICT Education @ DISI
DISI - Department of Information Engineering and Computer Science

# Topics 2001

## Projects in MicroElectronics and VLSI design

Contact person Dr. Mario Zen (IRST)
Contact person Giovanni Sonicini and Alessandro Zorat (University of Trento)
The area of microelectronics and microsystems is the common denominator for a number of projects that range from the design and implementation of architectures suitable for embeddable systems, to the design, fabrication, and testing of integrated circuits.
These projects are part of a research effort that is of interest to the IRST research institute, the university of Trento, and the local electronic industry.
General description of the projects:

• Design and fabrication of microsystems
• Silicon microdevices and electro-optical microsystems
• CMOS image sensors and processors

## Design and fabrication of microsystems

Contact person Dr. Mario Zen
At ITC-Irst Microsystems Division there are capabilities in simulation, design, fabrication and testing of microsystems. Three main areas of interest can be selected for possible joint collaborations. The first regards the development and production of double-sided large-area sensors for high energy physics experiments. On going is the development of fabrication technologies for double-sided, AC coupled microstrip detectors for ALICE tracker (LHC-Large Hadron Collider experiment at CERN) and the production of 400 double-sided microstrip detectors for the silicon tracker of AMS experiment (to be hosted in the Space Shuttle in 2003).
The second main field of interest is the consolidation and development of new technologies in the microsystem sector aimed at the production of prototypes of interest of leading Italian companies in a strategic vision whereby the Division devises and develops microsystems which are then transferred to companies for exploitation on an industrial scale. In particular we are developing technologies for biosensors of medical interest and microsystem for environmental monitoring as well as MEMS of specific industrial interest (pressure sensors, silicon membrane microphones, gas sensors, millimeter waves coplanar waveguides, . . .).
The third area concerns the development of integrated optical sensors, with particular emphasis to three-dimensional measuring systems. Smart optical ASICs for industrial applications in standard CMOS technology, design of mixed analog-digital circuits for optical sensor readout and pre-processing and integration of novel measuring techniques based on optical sensors are typical activities under development.
Specific project in this area are: OPTO-ASIC for high resolution absolute position sensor, high resolution, high speed self aligning bubble ink-jet actuator group, 2D-CMOS motion detection for security applications, advanced 1D-Linear Arrays for 3D measurements, 2D array for scannerless 3D vision sensor and development of CMOS digital cameras for automotive.

## Silicon microdevices and electro-optical microsystems

Contact person Giovanni Sonicini
Area of interest
MicroSystems (or MEMS/MOEMS: Micro Electro Optical Mechanical Systems) represent an emerging field with an increasing impact on a large variety of applications. MicroSystem technology draws its potential from a number of high-tech fields like Microelectronics, MicroOptics, MicroMechanics, thus enabling the development of innovative 3D miniaturized systems. New products based on MicroSystems technology are moving from the R&D laboratories into the market.
Proposed research activities
Preferably offered to graduate students interested in PhD fellowships.

#### Modelling of Silicon integrated microdevices for MEMS/MOEMS

Prototypes of Silicon based Microsystems of different type and of increasing complexity are beeig designed and fabricated at the Microfabrication Lab of IRST-MicroSystems Division in close cooperation with the ElectroOptical Lab of the Engineering Faculty of the University of Trento. In this development there is an increasing need of accurate modelling and numerical simulation of silicon integrated microdevices to be used in MicroSystems and of their fabrication process. Furthermore the mechanical, chemical and thermal properties of these microdevices as well as the design of dedicated packaging also requires an increasingly accurate modelling to reduce the costly "trial and error steps" on the way toward an optimized prototype. MicroSystems oriented Computer Aided Design (CAD) tools are already available on the market to assist the designer in optimizing the layout and the fabrication process of microdevices (both electrical, mechanical and optical) to be assembled into MicroSystems; however, due to the complexity and the novelty of MicroSystems technology, complete and reliable CAD for the predictive simulation of all the design level, from layout and processing to system operation, are still not available.
The proposed research activity in this field refers to a critical evaluation of the modelling of silicon integrated/micromachined test structures carried out with exhisting CAD tools as compared with experimental results. A variety of in-house already developed test structures, ranging from basic pressure sensors to radiation detectors, microheaters etc, as well as the possibility to develop new "to-the-purpose designed" test structures will allow an accurate analysis of the exhisting simulation tools capabilities and will suggest new and more accurate theoretical approaches to the mathematical simulation of microstructures.

#### ElectroOptical MicroSystems

Optoelectronic devices assembled onto micromachined Silicon Optical Benches (SiOB) to form complete ElectroOptical MicroSystems (or MOEMS) is becoming increasingly relevant for a wide spectrum of applications. The use of micromachined silicon substrates allows an efficient coupling of optical emitters (LED or LASER) and photorecivers with optical fibres, waveguides and interface integrated electronics. In this way a variety of miniaturized devices of interest for optical telecommunications as well as for other industrial applications, ranging from sensors to automotive and space applications.
The proposed research activity in this field refers to the development of micromachined silicon substrates for the efficient low cost coupling of optoelectronic devices with optical fibres and integrated interface electronics to form ElectroOptical MicroSystems for optical communications and for automotive applications. Furthermore innovative sensors for colours detection and optical nose based on gas sensitive photodiode arrays are now being developed in the frame of national/european research contracts. These research activities require the development and optimization of improved surface/bulk micromachining tecnologies as well as the design and fabrication of "to the purpose" designed SiOB. It also requires the development of special packaging and electrooptical testing procedures for measuring the performance of the electro optical microsystem prototypes.

## CMOS Image sensors and processors

Contact person Alessandro Zorat
This project stems from previous efforts that produced an image sensor that was developed specifically for computer vision. With the advent of the "invisible computers" that can be found in stand-alone electronic devices (for example: those to measure car speed and traffic volume) there is a growing need to include some highly specialized image processing capability directly in the chip that hosts the sensor.
With current fabrication technology it is possible to manufacture CMOS chips that contain areas that are sensitive to incident light, analogue circuits to amplify signals, and areas that implement the usual circuits that are found in digital processors, thus opening the possibility of designing truly integrated image sensors and processor.
An attractive possibility is to include in the optical sensor the circuits needed to compute some information that is relevant for the application at hand.
As an example, consider a circuit that computes the pixel coordinates of all those pixels whose value is within a given range. This would be useful for applications that call for optical recognition of objects or surface profiles.
Another example would be that of a sensor that incorporates the circuits for the early image adjustments, such as computing the gamma correction, or the compensation factors for lens aberrations or defects.

## Traffic Grooming in WDM networks

Contact person Prof. Roberto Battiti
Our laboratory is currently developing competitive heuristics for joining several low capacity channels into a single high-capacity wavelength running as a SONET or SDH circuit on an optical ring. The purpose is to minimize the hardware cost (Add-Drop multiplexers are very costly) at the various nodes.
The project has the purpose to develop C++ code in order to compare different algorithms for the problem and analyze experimental results.
Prerequisites: familiarity with the literature about the problem, C++ programming, Perl

## All-optical routing

The enormous speed of fiber optics is sensibly reduced by the need to convert optical signals into the electrical domain for various reasons: amplification, conversion, header inspection. New purely optical devices are under development: they will allow the major backbones to connect directly through assigned lightpaths. Most optimization problems, however, are difficult to solve and heuristic techniques are needed to get near-optimal solutions.
The project has the purpose to develop C++ code in order to compare different algorithms for the problem and analyze experimental results. Prerequisites: familiarity with the literature about the problem, C++ programming, Perl.

## Efficient Use of Radio Spectrum in Wireless Local Loop Networks with Channel Separation between Close Stations

We have recently studied the problem of assigning channels to the stations of a wireless network so that interfering transmitters are assigned channels with a given separation and the number of channels used is minimized. Two versions of the channel assignment problem have been considered which are equivalent to two specific coloring problems -- called L(2,1) and L(2,1,1) -- of the graph representing the network topology. In these problems, channels assigned to adjacent vertices must be at least 2 apart, while the same channel can be reused only at vertices whose distance is at least 3 or 4, respectively. Efficient channel assignment algorithms using the minimum number of channels have been provided for specific, but realistic, network topologies, including buses, rings, hexagonal grids, bidimensional grids, cellular grids, and complete binary trees. We are now on the way of expanding this study along three directions. First of all, we are willing to solve this problem for perturbed regular wireless local loop networks, that is, networks with regular topology -- like the ones we had already investigated -- with a given percentage of edges and nodes missing. Secondly, we would like to investigate this problem for arbitrary graphs. Finally, we plan to extend the solution to the case where not only a single channel but a set of constrained channels must be assigned to the nodes.

## A Column Classifier Algorithm

Recently a Column Classifier algorithm, has been proposed by Olariu and Pinotti which is able to classify a population of $r \times s$ item in $O(s)$ applications of a classifier device of size $r$. This algorithm leads to an exact partition of the input population. However, we are aware of several applications in digital signal processing, where to have almost a partition is more than enough. As a part of this project, we would like to make our Column Classifier algorithm weaker, avoiding to follow all the way down the recursion in it, in order to measure how far is the obtained partition from the exact one. This should lead to a probabilistic analysis of our algorithm.

## Intelligent retrieval of images in web environment using object shapes

Contact person Francesco De Natale, Fabrizio Granelli
In the framework of distributed multimedia information systems, a fundamental problem is to index and retrieve the information in an effective way. Text-based queries are often not efficient in locating multimedia data, due to the underlying ambiguity in the description of a visual or audio contents. Several activities have been undertaken within the Signal Processing and Telecommunications Group at the University of Trento to study the possibility of retrieving visual contents based on a set of descriptive features. The idea of the proposed project is to develop an innovative methodology to detect the most significant object shapes from an image, to be used in the context of shape-based queries. The technique should work directly in the compressed domain and should have a very low computational complexity. This activity is partially supported by the European Commission, within the framework of the IST R&D Programme (V F.P.), projects PSI3 and PERSEO.

## Change detection in multitemporal remote-sensing images

Contact person Lorenzo Bruzzone
In recent years, the detection and analysis of changes in multi-temporal remote-sensing images has assumed an ever increasing strategic role in several application domains. This is a consequence of the fact that a wide range of applications can benefit from the analysis of images acquired over a specific area of interest at different times, as proved by the impressive amount of recent literature dealing with the application of change analysis to different problems. Examples of these problems are monitoring of the pressure on the environment, monitoring of agricultural areas, global change analysis, and the assessment of damages due to forest fires, deforestation, floods, earthquakes, and volcanic activities. Several activities have been undertaken within the Remote Sensing Laboratory of the Signal Processing and Telecommunications Group at the University of Trento for developing accurate and effective automatic change-detection techniques. The proposed project concerns the development of a novel context-based method for detecting changes in multitemporal images. The objective of this method is to increase the accuracy of the change detection process by exploiting the spatial context information of images, without significantly degrade the precision in the location of borders of changed areas. The technique will be developed in the framework of a project supported by the Italian Space Agency.