Probabilistic Neural Networks Models for Accurate Atmospheric Modelling - Expro Plus

• Country: Netherlands
• Summary: Probabilistic Neural Networks Models for Accurate Atmospheric Modelling - Expro Plus
• NLT Ref No: 103517971
• Deadline: 23 Aug 2024
• Financier: Agency for the Cooperation of Energy Regulators (ACER)
• Purchaser Ownership: Government
• Tender Value: Refer Document
• Notice Type: Tender
• Document Ref. No.: 1-12196
• Purchaser's Detail:
  Login to see full purchaser details. Login to see full purchaser details. Login to see full purchaser details. Login to see full purchaser details. Login to see full purchaser details. Login to see full purchaser details. Login to see full purchaser details. Login to see full purchaser details. Login to see full purchaser details.

  Login to see details
• Description:
• Tenders are invited for Probabilistic Neural Networks Models for Accurate Atmospheric Modelling - Expro Plus The use of Neural Network (NN) approaches are being wide spread in the scientific community, theNN can capture a lot of information in a relatively simple forward model. However, the focus has sofar been put on the maximum fidelity of the resulting models neglecting the estimation of associatederrors. This may be acceptable for other fields but in many engineering applications - and speciallyin Navigation - the uncertainty is almost as important as the parameter determination. In this regard,several strategies are being put forward in the Machine Learning community, being the mostpromising the use of Bayesian NN, which belong to the Probabilistic NN (PNN) family.The activity will develop and test different PNN algorithms to determine an accurate troposphereand ionosphere delay together with the associated uncertainty. This could enable new capabilities,such as the provision of estimates of the model uncertainties, and a NN capable of providing globalatmosphere models to speed up convergence time of high-accuracy GNSS solutions with minimalbandwidth requirements.It is of interest to develop and test new strategies to use Probailistic NN (with different architecturesand types) to be able to understand the output uncertainty in them and its use for geophysicalestimation. This last part will be done by using the tropospheric and ionospheric delay estimates asa proof-of-concept that the uncertainties are indeed sufficiently accurate for Navigation applications(quasi-HAS convergence level and better PVT with broadcast-like models), through a test campaignwith different types of receivers. Additionally, the ionosphere model will support the analysis of thedirect use of sTEC, which in turn will eliminate the need for a mapping function, which is a sourceof problems for ionospheric modelling.This activityencompasses the following tasks:- State of the art of PNN.- Review of PNN algorithms suitable for Troposphere/Ionosphere delay determination- Development/training of the PNN- Test campaign with some selected devices to test convergence and accuracy.- Lessonslearned and transfer to other propagation fields.Software shall be delivered under an ESA Software Community LicenceProcurement Policy: C(3) = Activity restricted to SMEs RD Entities. For additional information please go to:http://www.esa.int/About_Us/Business_with_ESA/Small_and_Medium_Sized_Enterprises/Opportunities_for_SMEs/Procurement_policy_on_fair_access_for_SMEs_-_the_C1-C4_Clauses
• Documents:

 Tender Notice