Crowd-Based Learning of Spatial Fields for the Internet of Things: From Harvesting of Data to Inference

Arias-de-Reyna, Eva; Closas, Pau; Dardari, Davide; Djurić, Petar M.

doi:10.1109/msp.2018.2840156

Cited by 28 publications

(9 citation statements)

References 31 publications

(38 reference statements)

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In this regard, a powerful tool is represented by the theory of Gaussian processes (GPs), which is widely adopted in machine learning [ 53 , 54 ]. However, its adoption in the context of crowd mapping for spatial field estimations might still suffer from complexity and memory issues.…”

Section: Numerical Results: Examples Of Crowd Mapping and Proactivmentioning

confidence: 99%

Crowd-Based Cognitive Perception of the Physical World: Towards the Internet of Senses

Pasolini

Guerra

Guidi

et al. 2020

Sensors

Self Cite

View full text Add to dashboard Cite

This paper introduces a possible architecture and discusses the research directions for the realization of the Cognitive Perceptual Internet (CPI), which is enabled by the convergence of wired and wireless communications, traditional sensor networks, mobile crowd-sensing, and machine learning techniques. The CPI concept stems from the fact that mobile devices, such as smartphones and wearables, are becoming an outstanding mean for zero-effort world-sensing and digitalization thanks to their pervasive diffusion and the increasing number of embedded sensors. Data collected by such devices provide unprecedented insights into the physical world that can be inferred through cognitive processes, thus originating a digital sixth sense. In this paper, we describe how the Internet can behave like a sensing brain, thus evolving into the Internet of Senses, with network-based cognitive perception and action capabilities built upon mobile crowd-sensing mechanisms. The new concept of hyper-map is envisioned as an efficient geo-referenced repository of knowledge about the physical world. Such knowledge is acquired and augmented through heterogeneous sensors, multi-user cooperation and distributed learning mechanisms. Furthermore, we indicate the possibility to accommodate proactive sensors, in addition to common reactive sensors such as cameras, antennas, thermometers and inertial measurement units, by exploiting massive antenna arrays at millimeter-waves to enhance mobile terminals perception capabilities as well as the range of new applications. Finally, we distillate some insights about the challenges arising in the realization of the CPI, corroborated by preliminary results, and we depict a futuristic scenario where the proposed Internet of Senses becomes true.

show abstract

Section: Numerical Results: Examples Of Crowd Mapping and Proactivmentioning

confidence: 99%

Crowd-Based Cognitive Perception of the Physical World: Towards the Internet of Senses

Pasolini

Guerra

Guidi

et al. 2020

Sensors

Self Cite

View full text Add to dashboard Cite

show abstract

“…Pametni uređaji u sebi imaju ugrađene senzore za praćenje i merenje određenih fizičkih entiteta. Prikupljeni i obrađeni podaci dovode do određenih informacija koje mogu da utiču na svakodnevne aktivnosti kao i na unapređenje postojećih tehnologija (poznavanje propagacije radio talasa utiče na poboljšanje performansi radio sistema [1]). Lokalizacija i merenje svetlosnih karakteristika u nekom regionu od interesa su dve nerazdvojive i međusobno zavisne celine.…”

Section: Napomenaunclassified

“…Sa druge strane, čak i da je model posmatranog polja tačan, u slučajevima pogešne lokalizacije, njegovi rezultati su beskorisni. Same tehnike pozicioniranja često zavise od parametara modela i ovaj paradoksalni (chicken-and-egg) problem predstavlja osnov za razvoj metoda koje se bave združenim tehnikama lokalizacije i učenja prostornog polja [1].…”

Section: Slika 1 Robot I Pametni Telefon U Ulozi Agentaunclassified

Proračun Iluminacije Primenom Neuralnih Mreža

Latinović¹

2020

Zbornik radova FTN

View full text Add to dashboard Cite

U ovom radu opisana je primena LED svetla za potrebe sistema pametne rasvete, njegova alternativna uloga u sistemima za komunikaciju vidljivom svetlošću, VLC (Visible Light Communication) i značaj tih sistema u tehnikama pozicioniranja i lokalizacije. U radu su opisane metode estimacije iluminatornog kao prostornog polja, pri čemu je akcenat stavljen na unutrašnje prostore. Cilj ovog rada jeste da se na osnovu numeričkog eksperimenta dobijenog primenom alata programskog jezika MATLAB, za određenu prostoriju i definisane parametre LED rasvete, uradi procena iluminatornog polja u okviru odgovarajuće neuralne mreže.

show abstract

“…Alternatively, Federated Learning (FL), has recently attracted great interest due to its privacy-protecting nature and the efficient use of resources by harnessing the processing power of edge devices [3]. In the context of localization, a number of studies have shown the potential of distributed learning [4]- [6].…”

Section: Introduction and Related Workmentioning

confidence: 99%

Personalized Federated Learning over non-IID Data for Indoor Localization

Imbiriba

Park

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Localization and tracking of objects using datadriven methods is a popular topic due to the complexity in characterizing the physics of wireless channel propagation models. In these modeling approaches, data needs to be gathered to accurately train models, at the same time that user's privacy is maintained. An appealing scheme to cooperatively achieve these goals is known as Federated Learning (FL). A challenge in FL schemes is the presence of non-independent and identically distributed (non-IID) data, caused by unevenly exploration of different areas. In this paper, we consider the use of recent FL schemes to train a set of personalized models that are then optimally fused through Bayesian rules, which makes it appropriate in the context of indoor localization.

show abstract

Crowd-Based Learning of Spatial Fields for the Internet of Things: From Harvesting of Data to Inference

Cited by 28 publications

References 31 publications

Crowd-Based Cognitive Perception of the Physical World: Towards the Internet of Senses

Crowd-Based Cognitive Perception of the Physical World: Towards the Internet of Senses

Proračun Iluminacije Primenom Neuralnih Mreža

Personalized Federated Learning over non-IID Data for Indoor Localization

Contact Info

Product

Resources

About