Among positioning techniques ,as an alternative to analyzing sophisticated signal propagations, fingerprinting adopts a pattern matching approach. The main idea is to collect signal features of all possible locations in the area of interests to build a fingerprint database (known as site-survey or calibration). Localization is then simply the process of matching the measured fingerprints at an unknown location with those in the database and returning the location corresponding to the best-fitted fingerprint. We have chosen the RSSI (received signal strength indicator), the PDP (power delay profile) and the CIR (channel impulse response) as the ‘signature’ pattern, called fingerprint. In profiling based schemes, RSSI is considered as a quantity that depends on the distance between a transmitter and a receiver as well as on the indoor environment. Thus we may expect that the RSSI readings from similar environments may behave similarly. This hope lies behind the construction of a radio map of the monitored area by gathering the RSSI readings from known locations. The RSSI is captured through a set of infrastructure nodes (AP). To estimate the location of a target point q, based on a given set of RSSI readings ϕ, this map is explored to search for a set of nearest neighbors of ϕ,. In the radio map the locations of those chosen neighbors are also stored and they are used to predict the location of q. However, the accuracy performance suffers from the time-varying characteristic of RSSI due to the multipath and shadow effect. By considering the indoor environment, the multipath caused by reflections, diffraction and diffusion on the rough sidewall surfaces, and the non-line of sight (NLOS) due to the blockage of the shortest direct path between transmitter and receiver are the main sources of range measurement errors. In order to improve accuracy of localization we propose a fingerprinting positioning system based on power delay profile (PDP). The system extracts the PDP of the multipath from the cross-correlation result between the received signal and the reference signal. And to fully characterize the individual paths, we introduce the CIR(corresponding to the channel impulse response) as the fingerprint signature. This thesis aims to provide both theoretical and experimental contributions. It mainly studies fingerprint signatures for localization in indoor wireless sensor networks. It firstly introduce the wireless sensor network (WSN) indoor localization and the influence due to the indoor environment. Then different fingerprint signatures (RSSI,PDP,CIR) are proposed. And then the Matlab simulation of channel model and the numerical results are described in detail. Finally, the conclusion is drawn through the analysis of the numerical results. The structure of this thesis with the main contents of each chapter are reported below: Chapter 1 reports an overview of the wireless sensor network(WSN) indoor localization, its features and its advantages over a traditional wired system, and then compares the difference between indoor and outdoor environment. Finally,it describes the indoor positioning systems. Chapter 2 explains the signature in detail. It describes the indoor localization through fingerprinting (including off line and online procedures) and reports different signatures( RSSI,PDP,CIR ) used as fingerprints. It describes also how each signature works and its pros and cons Chapter 3 introduces the channel model and reports the a localization algorithms named Nearest Neighbor (NN) and Weighted K-Nearest Neighbor(WkNN). Explanations of the procedures and comparisons with kNN(K-nearest neighbor) are provided. Chapter 4 is about data analysis and presents the results drawn from the Matlab simulations.The Matlab simulations include the channel model, the format of RSSI, PDP, CIR and the algorithms WkNN,kNN,NN. Finally, the conclusions summarize the main results and the future work.
Nell’ambito della localizzazione, come alternativa all’analisi di sofisticati misure tratte dalla propagazione dei segnali, la tecnica di “fingerprinting” (impronte) adotta un approccio pattern matching. L'idea principale è quello di raccogliere le caratteristiche di tutte le posizioni possibili per la costruzione di una banca dati di impronte del segnale nell’area in questione. La localizzazione è quindi semplicemente il processo di corrispondenza delle impronte misurate in una posizione sconosciuta con quelli della banca dati con la stima della posizione corrispondente alla migliore impronta. Nella tesi, abbiamo scelto di utilizzare l’RSSI (intensità del segnale ricevuto), PDP (profilo di potenza in funzione del ritardo) CIR (risposta all’impulso del canale) come modelli di impronta. Nello schema di localizzazione, RSSI è considerato come una quantità che dipende dalla distanza tra un trasmettitore e un ricevitore, nonché dall'ambiente interno. Così possiamo aspettarci che le letture RSSI da ambienti simili possano comportarsi in modo simile. Questa ipotesi sta dietro la costruzione di una mappa radio dell'area monitorata raccogliendo le letture RSSI dalle posizioni note. L’RSSI viene poi misurato da un insieme di nodi dell'infrastruttura (AP). Per stimare la posizione di un punto q, sulla base di un dato insieme di RSSI letture φ, questa mappa è esplorata per la ricerca di una serie di possibili vicini di φ ,. Nella ricerca vengono selezionati anche più vicini che vengono utilizzati per prevedere la posizione di q. Tuttavia, le prestazioni in termini di precisione , soffrono la caratteristica variabile nel tempo di RSSI a causa del multi-percorso e dell’effetto ombra. Considerando un ambiente interno, la propagazione multi-percorso causata dalle riflessioni, la diffrazione, la diffusione sulle superfici laterali, e la condizione di assenza di percorso vista (NLOS) a causa del blocco del percorso diretto più breve tra trasmettitore e ricevitore sono le principali fonti di errore nella misura. Al fine di migliorare la precisione della localizzazione, proponiamo un sistema di posizionamento “fingerprinting” basato sul profilo di potenza-ritardo (PDP). Il sistema estrae la posizione dal confronto tra il PDP misurato nel segnale ricevuto e quello nei segnali del database. E per caratterizzare completamente i percorsi individuali, si introduce anche il CIR (corrispondente alla risposta all'impulso del canale) come proposta di impronta. Questa tesi si propone di fornire contributi sia teorici che sperimentali. Vengono studiate principalmente le impronte per la localizzazione in reti di sensori wireless indoor. Poi sono proposte e simulate diverse impronte (RSSI, PDP, CIR). La struttura di questa tesi con i principali contenuti di ogni capitolo sono riportati di seguito: Il capitolo 1 riporta una panoramica delle reti wireless di sensori (WSN) e della localizzazione indoor, le sue caratteristiche ed i suoi vantaggi rispetto ad un sistema cablato tradizionale, e quindi confronta la differenza tra un ambiente interno e uno esterno. Infine, descrive il sistema di posizionamento indoor. Il capitolo 2 spiega il principio del “fingerprinting” in dettaglio. Descrive la localizzazione indoor tramite impronte (comprese le procedure off line e on-line) e le relazioni tra diverse impronte (RSSI, PDP, CIR) Viene poi descritto come ogni impronta funziona e i suoi cantaggi e svantaggi. Il capitolo 3 introduce il modello di canale e descrive l’algoritmo di localizzazione basato sul più vicino (NN) o il (WkNN). Le spiegazioni degli algoritmi sono completate da un confronto sulle prestazioni Il capitolo 4 presenta i risultati tratti dalle simulazioni Matlab. In Matlab, abbiamo costruito il modello di canale, e gli algoritmi basati su RSSI, PDP, CIR. La tesi si conclude con le osservazioni su possibili sviluppi futuri.
Fingerprinting signatures for localization in indoor wireless sensor networks
JIAO, XIN
2015/2016
Abstract
Among positioning techniques ,as an alternative to analyzing sophisticated signal propagations, fingerprinting adopts a pattern matching approach. The main idea is to collect signal features of all possible locations in the area of interests to build a fingerprint database (known as site-survey or calibration). Localization is then simply the process of matching the measured fingerprints at an unknown location with those in the database and returning the location corresponding to the best-fitted fingerprint. We have chosen the RSSI (received signal strength indicator), the PDP (power delay profile) and the CIR (channel impulse response) as the ‘signature’ pattern, called fingerprint. In profiling based schemes, RSSI is considered as a quantity that depends on the distance between a transmitter and a receiver as well as on the indoor environment. Thus we may expect that the RSSI readings from similar environments may behave similarly. This hope lies behind the construction of a radio map of the monitored area by gathering the RSSI readings from known locations. The RSSI is captured through a set of infrastructure nodes (AP). To estimate the location of a target point q, based on a given set of RSSI readings ϕ, this map is explored to search for a set of nearest neighbors of ϕ,. In the radio map the locations of those chosen neighbors are also stored and they are used to predict the location of q. However, the accuracy performance suffers from the time-varying characteristic of RSSI due to the multipath and shadow effect. By considering the indoor environment, the multipath caused by reflections, diffraction and diffusion on the rough sidewall surfaces, and the non-line of sight (NLOS) due to the blockage of the shortest direct path between transmitter and receiver are the main sources of range measurement errors. In order to improve accuracy of localization we propose a fingerprinting positioning system based on power delay profile (PDP). The system extracts the PDP of the multipath from the cross-correlation result between the received signal and the reference signal. And to fully characterize the individual paths, we introduce the CIR(corresponding to the channel impulse response) as the fingerprint signature. This thesis aims to provide both theoretical and experimental contributions. It mainly studies fingerprint signatures for localization in indoor wireless sensor networks. It firstly introduce the wireless sensor network (WSN) indoor localization and the influence due to the indoor environment. Then different fingerprint signatures (RSSI,PDP,CIR) are proposed. And then the Matlab simulation of channel model and the numerical results are described in detail. Finally, the conclusion is drawn through the analysis of the numerical results. The structure of this thesis with the main contents of each chapter are reported below: Chapter 1 reports an overview of the wireless sensor network(WSN) indoor localization, its features and its advantages over a traditional wired system, and then compares the difference between indoor and outdoor environment. Finally,it describes the indoor positioning systems. Chapter 2 explains the signature in detail. It describes the indoor localization through fingerprinting (including off line and online procedures) and reports different signatures( RSSI,PDP,CIR ) used as fingerprints. It describes also how each signature works and its pros and cons Chapter 3 introduces the channel model and reports the a localization algorithms named Nearest Neighbor (NN) and Weighted K-Nearest Neighbor(WkNN). Explanations of the procedures and comparisons with kNN(K-nearest neighbor) are provided. Chapter 4 is about data analysis and presents the results drawn from the Matlab simulations.The Matlab simulations include the channel model, the format of RSSI, PDP, CIR and the algorithms WkNN,kNN,NN. Finally, the conclusions summarize the main results and the future work.File | Dimensione | Formato | |
---|---|---|---|
2016_07_Jiao.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Testo della tesina
Dimensione
1.68 MB
Formato
Adobe PDF
|
1.68 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/123579