←
Ben Thomas80 posts
Ben Thomas avatar

Ben Thomas

@BenBillThomas

Research associate @UniofBath. Interested in underwater navigation and remote sensing with a focus on interferometric synthetic aperture sonar.

◷ Joined May 2018▧ 19 Photos and videos
Ben Thomas avatar
Ben Thomas@BenBillThomas·2 Feb 2022

Looking forward to hosting the next @acoustics_ac_uk @sigunderwater webinar! We'll be hearing about underwater noise modelling and passive acoustic monitoring from industrial and governance perspectives.

The next Underwater Acoustics webinar is on 24th February at 2pm. Presenting are: Rasmus Pedersen: Practical Underwater Noise Modelling for the Offshore Industry Monika Kosecka: The Potential of Passive Acoustic Monitoring in Fish Studies Register via: acoustics.ac.uk/underwater-a…

Post attachment
Ben Thomas avatar
Ben Thomas@BenBillThomas·6 Dec 2021

Here's my attempt at an example submission for the @acoustics_ac_uk @sigunderwater 'Research Soundings' early career competition. You have until 10 December!

0:30

The deadline has been extended to the 10th December for our Research Soundings early career competition! Submit your 30 second audio clip along with up to 150 words of description to be in with a chance of winning a share of the £1000 prize money! acoustics.ac.uk/research-sou…

Alan Hunter avatar
⟳ Ben Thomas retweeted
Alan Hunter@DrAlanHunter·11 Nov 2021

"A picture might be worth a thousand words but a sound could be worth a thousand pounds!": submit a 30s audio clip relating to your research for the @sigunderwater @acoustics_ac_uk PhD Symposium and you could win a prize of £500, £300, or £200. Please retweet.

Competition time! Submit 30 seconds of sound about your work along with up to 150 words to be in with a shot of winning up to £500! acoustics.ac.uk/research-sou…

Post attachment
Ben Thomas avatar
Ben Thomas@BenBillThomas·8 Oct 2021

Just moved the last of my data off the @BathHPC Balena system before its impending retirement. It has served us well! I have been trying out the new @Azure cloud-based system this week and it seems great... I just hope I am this positive once the billing system is implemented!

Ben Thomas avatar
Ben Thomas@BenBillThomas·28 Sep 2021

Join us this Thursday @ 2pm for the @acoustics_ac_uk @sigunderwater monthly webinar! We're talking ocean observation using acoustics, gliders, signal processing and its implementation on FPGAs.

UKAN+ Underwater Acoustics Monthly Webinars are back! On Thursday 30th September at 2-3pm, we'll be discussing signal processing for underwater acoustic monitoring from low-power platforms e.g. gliders. More info at acoustics.ac.uk/underwater-a… Register at: zoom.us/webinar/register/WN_…

Post attachment
Ben Thomas avatar
Ben Thomas@BenBillThomas·24 Sep 2021

These developments represent an important step towards fusing micronavigation estimates into autonomous underwater vehicle navigation systems, which will enable higher accuracy underwater navigation, allowing longer underwater transits and improved geolocation of seafloor images.

Ben Thomas avatar
Ben Thomas@BenBillThomas·24 Sep 2021

Using simulated synthetic aperture sonar data, we demonstrate that the bias mitigation strategies we describe result in reductions in along-track micronavigation bias and standard deviation of 48.5%–99.5% and up to 34% respectively. 👇

Ben Thomas avatar
Ben Thomas@BenBillThomas·24 Sep 2021

While a Gaussian interpolation kernel fits the along-track spatial coherence function well, we derive an improved interpolation kernel for rectangularly shaded apertures with sinc beampatterns, which is a cubic function. The plots show their respective kernel fit errors. 👇

Post attachmentPost attachment
Ben Thomas avatar
Ben Thomas@BenBillThomas·24 Sep 2021

The third bias is caused by a mismatch between the interpolation kernel and the underlying coherence function, which looks like this as a function of along-track offset and range. 👇

Post attachment
Ben Thomas avatar
Ben Thomas@BenBillThomas·24 Sep 2021

In the article, we time-align the echo signals before estimating the sample coherence, ensuring they are valid spatial coherence estimates. Also, we introduce a Bayesian estimator across the diagonals of the grid to mitigate the bias of sample coherence estimates. 👇

Post attachment
Ben Thomas avatar
Ben Thomas@BenBillThomas·24 Sep 2021

Unfortunately, the sample coherence estimator tends to over-estimate the underlying coherence. Also, the signals must be temporally registered for the coherence estimator to be a valid estimator of spatial coherence, that is, how the coherence changes with spatial offset. 👇

Ben Thomas avatar
Ben Thomas@BenBillThomas·24 Sep 2021

Conventionally, we take the mean coherence across the diagonals and fit a Gaussian interpolation kernel through the points. The location of the peak of this function is the estimate of the along-track offset, and the inter-ping distance travelled follows from this. 👇

Post attachment