Transfer Learning for Convolutional Indoor Positioning Systems
Creators
- 1. Tampere University, Finland
- 2. Tampere University, Finland; Universitat Jaume I, Spain
- 3. UBIK Geospatial Solutions S.L., Spain
Description
Fingerprinting is a widely used technique in indoor positioning, mainly due to its simplicity. Usually, this technique is used with the deterministic k - Nearest Neighbors (k-NN) algorithm. Utilizing a neural network model for fingerprinting positioning purposes can greatly improve the prediction speed compared to the k-NN approach, but requires a voluminous training dataset to achieve comparable performance. In many indoor positioning datasets, the number of samples is only at a level of hundreds, which results in poor performance of the neural network solution. In this work, we develop a novel algorithm based on a transfer learning approach, which combines samples from 15 different Wi-Fi RSS indoor positioning datasets, to train a single convolutional neural network model, which learns the common patterns in the combined data. The proposed model is then fine-tuned to optimally fit the individual databases. We show that the proposed solution reduces the positioning error by up to 25% compared to the benchmark model while reducing the number of outlier predictions.
Notes
Files
IPIN_2021_Transfer Learning_FRONT PAGE.pdf
Files
(448.3 kB)
Name | Size | Download all |
---|---|---|
md5:667aed6296e958c0add42e5ed96acc7b
|
448.3 kB | Preview Download |
Additional details
Funding
- A-WEAR – A network for dynamic WEarable Applications with pRivacy constraints 813278
- European Commission
- High-Efficiency Localization and Location-aware Communications in Future Mobile Networks 323244
- Academy of Finland
- Ubiquitous Localization, communication, and sensing infrastrucTuRe for Autonomous systems (ULTRA) / Consortium: ULTRA 328214
- Academy of Finland
- Research Infrastructure for Future Wireless Communication Networks / Consortium: FUWIRI 319994
- Academy of Finland
- Research Infrastructure for Future Wireless Communication Networks / Consortium: FUWIRI-6G 338224
- Academy of Finland