Optical feedback interferometry (OFI)

Such devices prove their interest in terms of sensitivity and resolution. OFI allows the laser diode to be used as a stand-alone micro-interferometer, i.e. by incorporating the light source and the interferometer in the active cavity of the laser itself without external optical components.

OFI sensing basic set-up. Compared to usual interferometry, there is no external optical component.

Principe général d’un dispositif de mesure par IRO. On note l’absence de composants optiques externes comparativement à de l’interférométrie classique.

However, they still offer nowadays a very limited versatility in their range of applicability. Even minor changes on the system specifications (bandwidth, range of measurements, accuracy...) imply usually a complete redesign of the device.

Our strategy is then to provide relaxed constraints in the design of our embedded systems for a variety of requirements with the specificity of avoiding external optical components contrary to the usually proposed solutions.



Performances of our sensing devices are first ameliorated by modelling the physical behaviour of OFI signals and then using a modular approach prone to hardware-software partitioning of our robust real-time signal processing algorithms with a specifically designed electronics.

By considering the relaxation dynamics of a long external cavity when several solutions to the steady-state equations can be obtained and a change in the laser-to-target distance can lead to a bifurcation in the set of solutions inducing a discontinuity in the modelled power of a laser diode, we demonstrated that high-frequency damped oscillations appear with a high-bandwidth acquisition electronics (100 MHz) instead of usual discontinuities. This transient phenomenon in a single fringe contain information about the target reflectivity r3 and absolute distance (i.e. time-of-flight) ext to the target paving the way to a brand new generation of OFI sensors.


Frange d’IRO mesurée avec un détecteur large bande passante (100MHz).

Solid: measured time series of a typical signal for a medium r3  (τext = 14.9 ns, averaging on 512 acquisitions).
Dashed: the same signal, filtered. The usual sawtoothlike signal can be observed.

Simulated signal in the dynamical model for two different  r3 and τext
Signaux simulés à l’aide de notre modèle dynamique pour deux valeurs de r3 et τext

Critical applications

New associated designs are then required in OFI for critical applications such like: mechatronics (real-time vibration analysis for non-destructive testing, quality control), fluidics (sustainable energy, biomedical sensing) and detonics.


 We have proposed a model to achieve subnanometric accuracy for displacement measurements before developing a new Phase Unwrapping Methods (PUM) generally limited to a 20 nm precision due to misinterpreted signal phase around OFI signal discontinuities. Using coupled delayed differential equations, we ensure that a truly bijective function between OFI signals and phase can be defined, thereby allowing to reach subnanometer precisions.

Classification of the moderate OFI signal in 6 sub-classes function of the shape of the signal.

As the coupling target-laser feedback factor C is continuously changing the OFI signal shapes, it is also a major cause of accuracy loss. We proposed a global optimization performed on only a fraction of the collected sample to follow the evolution of C. Real-time measurement of C has been developed to perform autofocus on the target with adaptive optics.

By performing data fusion (phase and gain) with a solid-state accelerometer (SSA)to correct the influence of parasitic movement ot the OFI sensor itself, we have patented and designed the very first embedded OFI vibrometer. Even when subject to extraneous movements, our vibrometer provides accurate measurements (40 nm) limited by the accuracy of the SSA itself.

Block diagram of the Solid-State Accelerometer coupled Self-Mixing (SSA-SM) sensor:
photodiode (PD), laser diode (LD), collimating lens (CL), and piezoelectric transducer (PZT).
« European Mechatronics Awards 2010 », Research category.

Signal carried out by the laser device subject to parasitic vibrations for a motionless target.
An improvement of 31dB or 96% elimination of the parasitic vibration acting on prototype is achieved.

Displacement reconstruction with the capacitive sensor inside the PZT (in green)
and with our OFI  vibrometer (in red). In blue, OFI variations of the optical output power.

Including a real-time solution based on Hilbert Transform to remove the influence of the speckle effect, we can reconstruct measurements in the time domain - even in case of strong fading of the amplitude of the OFI signal - with all sorts of non-cooperative targets from very rough (sandpaper) to dark (back side of a mouse carpet).

Lastly, we have coupled the vibrometer with optical fibers for studying dynamics of material subject to an explosion (pressure 100 GPa, rising time 0.5 µs, stored energy 70kJ, load voltage 85kV). We have measured during the first 5 µs a displacement of a target presenting a velocity of 2000 km/h with a 2 GHz photodiode.


The laser is directly used as a probe in a complex heterogeneous medium (like the skin).

Experimental basic set-up for flow measurements.

A photon is not reflected by a single particle in the flow but suffers from multiple reflections before being re-injected into the laser cavity. The obtained signal includes a whole spectrum of frequencies (spread Doppler frequency shift) which is rather complex to monitor. Moreover as the amount of power back-scattered into the laser cavity by blood cells is low, it has been necessary to extract from the OFI model the parameters that increase its signal-to-noise ratio.

Spread Doppler frequency shift obtained by our OFI velocimeter for different concentration of diluted milk

OFI signal in the frequency domain for several flow rates (0, 10, 20, 30, 40 and 50 μL.min−1)

Our goal is first to investigate turbulent flows in microfluidics devices and more particularly helping noninvasive early detection of melanomatous skin cancer with rapid population screening and office diagnosis.

We have ameliorated the spatial resolution of our OFI velocimeter with a two-lenses set-up optimizing the collection of the backscattered light to increase the local power density. Temporal resolution has been increased by dedicated signal processing notably with a fully original architecture of frequencymeter coupled to auto-correlation, instead of time-consuming usual FFT.

Our flowmeter is able to achieve flow measurement in a microscale channel with dimensions down to 20µm and for a velocity range from 2µl/min to more than100µL/min (upper limitation due to the mechanical constraint on the channel).

Real-time reconstruction of velocity profiles in micro-channels (down to 20µm) is demonstrated in different configurations (fluid viscosity, scatterers density, micro-channel section type…). Our results have been compared advantageously with more mature technique such as dual-slit requiring at least 40s for one measured point and presenting several unaccurate measurements.

We are members of the COST Action BM1205: European Network for Skin Cancer Detection using Laser Imaging (http://skin-laser-imaging.org/)

Newtonian theoretical profile normalized by the maximal measured velocity
 (in green) in a 120 µm channel. Experimental results obtained from
the dual-slit technique (in red, 40s for one measured point).
Real-time measurements with the OFI velocimeter (in blue).


Detecting very low concentration of particles like aerosols carried by fluid medium is a challenge notably for sustainable energy like windmill.

We have designed an anemometer for measuring wind particles velocity.
It detects a single particle of small diameter in the Mie diffusion regime where the ratio of emitted power to the back-scattered power can reach 109.

Developed with a French start-up (http://www.epsiline.com/en), actual performances are a correlation of this anemometer measurement with classical anemometer better than 99%, for wind velocity up to 30 m/s implying signal acquisition and processing with frequency up to 100MHz at distances up to 4 m.

Prototype of the anemometer.