Signalbehandling

 

Forskargruppsledare: ​Professor Tomas McKelvey

Alla medarbetare i forskargruppen listas längre ner på sidan.

Om forskningsområdet Signalbehandling

Idag finns Digital Signalbehandling (DSP) i mobiltelefoner, CD-spelare, bilar, tvättmaskiner etc. Kort sagt, DSP finns överallt. Vad DSP oftast gör är att omvandla råa signaler från sensorer till användbar information. Ett mål är att ersätta dyra sensorer med billiga kombinerat med smart signalbehandling. Ett annat mål är att kunna mäta sådant som förut var omätbart.

Vi arbetar med forskning och utbildning i signaler och system samt digital signalbehandling. Vi har en stark bakgrund i modellbaserad signalbehandling, som använder verktyg från bland annat matematisk statistik och numeriska metoder. Våra ”klassiska” tillämpningar kommer från mobilkommunikation och radarsignalbehandling, där vi bland annat arbetar med kanalestimering, adaptiva antenner och högupplösande metoder. Vi har också som mål att sprida kunskapen till andra områden, där modern signalbehandling kan användas för att bryta ny mark. Exempel på sådana projekt inkluderar analys av signaler inom elkraftkvalitet, minspaning, bilelektronik (diagnos och styrning av motorer och aktiv säkerhet), objektdetektering i videokommunikation, samt analys av EEG och andra biomedicinska signaler.
 
Vår mix av grundforskning inom statistisk signalbehandling och nya tillämpningsområden har varit framgångsrik och vi kommer att fortsätta i samma anda. Vårt starka fokus på doktorandutbildning kommer också att fortsätta.

Forskningsprojekt

 
> Closed-loop control of combustion engines
A vital part of a system for closed-loop combustion control is sensors providing combustion information that can be fed back to the controller. This project investigates the use of a crankshaft integrated torque sensor for this purpose. The work includes development of methods for both combustion property estimation and closed-loop combustion property control.
 
> Light-Duty Diesel Engine for 2012 - Engine Control
This project investigates the use of information from crankshaft torque and ion current measurements for closed-loop diesel engine control. A large portion of the work therefore consists of developing methods for extracting combustion information from the measured signals.
 
​Projektet har två huvudsakliga arbetspaket: Etapp ett utgår ifrån projektet SEFS, och syftar till att leverera högpresterande modeller och metoder att användas i följningsfilter. En viktig frågeställning är om vi smidigt kan anpassa ett traditionellt följningsfilter så att det kan användas effektivt i etapp två...
 
> Multi-Antenna Technologies for Wireless Access and Backhaul (MATWAB)
In this project, we investigate the potential of two emerging technologies in the area of wireless communications. More specifically, we explore the concept of large-scale (or massive) MIMO systems which are anticipated to deploy base-stations with hundreds of low-power antennas.
 
This project concerns analysis and development of new algorithms for target tracking in complicated environments, involving multiple targets and uncertain measurements due to partial occlusion, heavy clutter, etc.
 
> Microwave tomography for pharmaceutical processes
This project aims at developing a sensor array of antennas capable of microwave tomography for pharmaceutical processes, together with a post-processing algorithm for data analysis.
 
 
> Non hit car and trucks
The project is focusing on jointly developing technologies to reduce accident risks for both passenger cars and commercial vehicles and particularly addressing the situations at which today’s active safety systems are not yet sufficient. To reach the goals brand new and improved safety functions with real-life benefits need to be invented across the whole safety domain, ranging from strategic drive to in-crash activities.
 
>  Object tracking for video surveillance and traffic safety
Research in this project mainly includes: video object tracking captured by one moving camera, from multiple camera, captured by visual and IR cameras. The primary methods in our study include, for example: particle filters, mean shift, local feature points, differentiable manifolds (e.g. Grassmann, Riemannian), online domain-shift learning, multiview geometry, and sensor fusion. In terms of problem solving, we focus on tracking dynamic objects through complex scenes containing long-term partial/full occlusions, large objects with significant out-of-plane pose changes. We track generic objects, e.g. humans, faces, vehicles, animals, workshop tools, and many more…
 
> Automatic classification and diagnostics of power system disturbance recordings
The aim is to automatic classification of the underlining causes of power system disturbances from a large amount of measurement data. This includes analyzing and characterizing power system disturbances, searching for common phenomena behind each type of underlying causes, and give recommendations.
 
> Human activity and behavior analysis and classification with applications
Research in this project mainly includes: analysis of a range of activities and behaviors (e.g. in elderly care centers, out-patient care centers, office environments, and vehicle drivers). This includes the recognition of different (or, specific) activities, analysis of individual activities to obtain a range of long-term (or short-term) statistics on normal activity/behaviors for purposes such as improving elderly care, detecting abnormality, office environment improvement, and safety driving. Main methods we investigate in this project are: novel machine learning and pattern classification methods, effective feature descriptors for activities/behaviors, statistical modeling and parameter estimations for normal/abnormal activities, and generate recommendations or trigger actions...
 
Lesions affecting the visual pathways in the human brain are common and may cause reduced visual acuity or visual field defects, either directly or as a result of surgery. These pathways can be visualised using tractography. The procedure is based on a combination of a magnetic resonance imaging technique known as diffusion tensor imaging (DTI) and computer-based image analysis.
 
> Parameter estimation based on sparse modeling
In this project, we focus on two particular problems related to model uncertainty. One concerns calibration using interpolation and smoothing, and a particular application is antenna array signal processing. The other is tracking of moving targets using multiple motion models.
 
 
 
 
> Optimization of sensors and sensor arrays
In this project the aim is to investigate how the robustness and accuracy of a sensor system is influenced by the number of sensors, their frequency range of operation and their specific locations around the measurement region.
 
 

Sortera efter

Avdelningar

Publicerad: fr 07 sep 2012. Ändrad: ti 21 nov 2017