This thesis is part of the AlphaSat experiment within the SCIEX (Satellite Communication Investigation and Experiment) campaign and focuses on the study and analysis of the satellite communication link between the AlphaSat satellite and the ground stations located in Tito Scalo and Spino d’Adda. In particular, the study concerns communications in the Ka and Q bands, which are strongly affected by atmospheric phenomena such as rain attenuation and tropospheric scintillation, representing significant sources of signal quality degradation. These two phenomena manifest themselves in the communication link with different spectral characteristics and require appropriate separation to be analyzed individually. The objective of this work was to develop a model capable of accurately and effectively representing the Power Spectral Density (PSD) of the observed events, enabling the separation of the spectral regions associated with rain and scintillation. To this end, an algorithm was implemented in MATLAB® to generate best-fit lines that adapt to the different spectral configurations observed experimentally while maintaining consistency with the theoretical reference model. In particular, the model imposes the theoretical constraints of characteristic slopes equal to -20 dB/dec for the rain-dominated region, 0 dB/dec for the transition (flat or plateau) tropospheric scintillation-dominated region, and −80/3 dB/decade for the decreasing tropospheric scintillation-dominated region. The construction of the model involves the selection of suitable frequency intervals within the three spectral regions and the computation, for each interval, of a best-fit line with the assigned theoretical slope. The extension and intersection of these lines made it possible to obtain a complete PSD model adaptable to the different spectral configurations observed experimentally. From this parametric representation of the spectrum, it was possible to automatically determine both the transition frequency associated with the rain component, used for the design of the low-pass filter, and the scintillation knee frequency, required for the definition of the band-pass filter, namely the key parameters for the separation of the rain and scintillation components. The model performance was evaluated by comparing the histograms of the automatically estimated characteristic frequencies with those of the manually identified frequencies, showing a good agreement between the two distributions. Furthermore, the Absolute Knee Frequency Error and the Absolute Rain Transition Frequency Error exhibited values close to zero, highlighting the high accuracy of the proposed method. The validity of the proposed model was assessed through a statistical analysis carried out on both a representative sample dataset and the complete database of events acquired at the Tito Scalo and Spino d’Adda ground stations. The consistency of the model was further assessed by comparing the crosswind velocities estimated from the knee frequencies identified by the algorithm in the Ka and Q bands. The analysis showed good agreement between the two estimates, with generally limited errors, confirming the overall coherence of the proposed method. The differences observed between the two analyzed stations are consistent with the performance of the models adopted for the estimation of the knee frequency.The analysis also revealed significant correlations between the characteristic frequencies and the attenuation intensity, especially the mean attenuation, enabling a better characterization of the spectral behavior of the events. Finally, the identified frequencies were used to implement an adaptive filtering procedure aimed at the automatic separation of rain attenuation and tropospheric scintillation components. The visual comparison between the total measured attenuation and the two components obtained from the filtering process showed a coherent reconstruction of the original signal, with an almost complete overlap between the measured signal and the sum of the separated components. This result provides additional confirmation of the correctness of the cutoff frequencies identified by the model and of the physical consistency of the adopted separation procedure. The obtained results demonstrate the validity of the proposed approach and its capability to correctly identify the spectral characteristics of the analyzed events. As future developments, further refinement of the frequency interval selection used for model construction is envisaged, together with the investigation of dedicated strategies for handling particular spectral configurations that do not exhibit the typical behavior observed in most of the analyzed events.
Questa tesi si inserisce nell'ambito dell'esperimento AlphaSat della campagna SCIEX (Satellite Communication Investigation and Experiment) ed è finalizzato allo studio e analisi della comunicazione satellitare tra il satellite AlphaSat e le stazioni di terra di Tito Scalo e Spino d’Adda. In particolare, lo studio riguarda i collegamenti nelle bande Ka e Q, fortemente influenzati da fenomeni atmosferici quali l'attenuazione dovuta alla pioggia e la scintillazione troposferica, che rappresentano importanti cause di degradazione della qualità del segnale. Questi due fenomeni si manifestano nella comunicazione con caratteristiche spettrali differenti e richiedono un’opportuna separazione per poter essere analizzati individualmente. L'obiettivo del lavoro è stato sviluppare un modello in grado di rappresentare in modo accurato ed efficace lo spettro di densità di potenza (Power Spectral Density, PSD) degli eventi osservati, consentendo la separazione delle regioni spettrali associate alla pioggia e alla scintillazione, in modo automatico. A tale scopo è stato implementato un algoritmo in ambiente MATLAB® che genera linee di best-fit aderenti alle diverse configurazioni spettrali riscontrate sperimentalmente, mantenendo al contempo la coerenza con il modello teorico di riferimento. In particolare, il modello impone i vincoli teorici delle pendenze caratteristiche pari a -20 dB/dec per la regione dominata dalla pioggia, 0 dB/dec per la regione di transizione (plateau o piatta) dominata dalla scintillazione e -80/3 dB/dec per la regione decrescente della scintillazione troposferica. La costruzione del modello prevede la selezione di opportuni intervalli di frequenza all'interno delle tre regioni dello spettro e calcolando, per ciascuno di essi, una retta di best-fit con la pendenza teorica assegnata. L'estensione e l'intersezione di tali rette hanno permesso di ottenere un modello completo della PSD, adattabile alle diverse configurazioni spettrali osservate sperimentalmente. Da questa rappresentazione parametrica dello spettro è stato possibile determinare automaticamente sia la frequenza di transizione associata alla componente di pioggia, utilizzata per la progettazione del filtro passa-basso, sia la frequenza di ginocchio della scintillazione, necessaria per la definizione del filtro passa-banda, ovvero i parametri fondamentali per la separazione delle componenti di pioggia e scintillazione. Le prestazioni del modello sono state valutate confrontando gli istogrammi delle frequenze caratteristiche stimate automaticamente con quelli delle frequenze identificate manualmente che confermano una buona concordanza tra le due distribuzioni e mediante i parametri di errore assoluto della frequenza del ginocchio ed errore assoluto della frequenza di transizione che mostrano valori prossimi allo zero, evidenziando un’elevata accuratezza del metodo. La validità del modello è stata verificata mediante un’analisi statistica condotta sia su un dataset campione sia sull’intero database degli eventi acquisiti nelle stazioni di Tito Scalo e Spino d’Adda. La consistenza del modello è stata inoltre verificata mediante il confronto delle velocità del vento trasverso stimate a partire dalle frequenze di ginocchio, trovate tramite l’algoritmo, nelle bande Ka e Q. L’analisi ha mostrato una buona concordanza tra le due stime, con errori generalmente contenuti, confermando la coerenza complessiva del metodo proposto. Le differenze osservate tra le due stazioni analizzate risultano coerenti con le prestazioni dei modelli adottati per la stima della frequenza di ginocchio. L’analisi ha inoltre evidenziato correlazioni significative tra le frequenze caratteristiche e l’intensità dell’attenuazione, soprattutto media, permettendo di caratterizzare il comportamento spettrale degli eventi. Le frequenze individuate sono state infine utilizzate per implementare una procedura di filtraggio adattiva, finalizzata alla separazione automatica delle componenti di pioggia e scintillazione troposferica. Il confronto visivo tra l’attenuazione totale misurata e le due componenti ottenute dal filtraggio ha evidenziato una ricostruzione coerente del segnale originale, con una sovrapposizione pressoché completa tra il segnale misurato e la somma delle componenti separate. Tale risultato fornisce un’ulteriore conferma della correttezza delle frequenze di taglio individuate dal modello e della consistenza fisica della procedura di separazione adottata. I risultati dimostrano quindi la validità dell'approccio proposto e la sua capacità di identificare correttamente le caratteristiche spettrali degli eventi analizzati. Come sviluppi futuri, si prevede la raffinazione della selezione degli intervalli di frequenza utilizzati per la costruzione del modello e di approfondire il trattamento e le strategie dedicate alla gestione di particolari configurazioni spettrali che presentano caratteristiche non riconducibili al comportamento tipico osservato nella maggior parte degli eventi.
Separazione automatica degli effetti della pioggia e della turbolenza troposferica nel segnale beacon Alphasat in banda Ka e Q
Malentacchi, Giulia
2025/2026
Abstract
This thesis is part of the AlphaSat experiment within the SCIEX (Satellite Communication Investigation and Experiment) campaign and focuses on the study and analysis of the satellite communication link between the AlphaSat satellite and the ground stations located in Tito Scalo and Spino d’Adda. In particular, the study concerns communications in the Ka and Q bands, which are strongly affected by atmospheric phenomena such as rain attenuation and tropospheric scintillation, representing significant sources of signal quality degradation. These two phenomena manifest themselves in the communication link with different spectral characteristics and require appropriate separation to be analyzed individually. The objective of this work was to develop a model capable of accurately and effectively representing the Power Spectral Density (PSD) of the observed events, enabling the separation of the spectral regions associated with rain and scintillation. To this end, an algorithm was implemented in MATLAB® to generate best-fit lines that adapt to the different spectral configurations observed experimentally while maintaining consistency with the theoretical reference model. In particular, the model imposes the theoretical constraints of characteristic slopes equal to -20 dB/dec for the rain-dominated region, 0 dB/dec for the transition (flat or plateau) tropospheric scintillation-dominated region, and −80/3 dB/decade for the decreasing tropospheric scintillation-dominated region. The construction of the model involves the selection of suitable frequency intervals within the three spectral regions and the computation, for each interval, of a best-fit line with the assigned theoretical slope. The extension and intersection of these lines made it possible to obtain a complete PSD model adaptable to the different spectral configurations observed experimentally. From this parametric representation of the spectrum, it was possible to automatically determine both the transition frequency associated with the rain component, used for the design of the low-pass filter, and the scintillation knee frequency, required for the definition of the band-pass filter, namely the key parameters for the separation of the rain and scintillation components. The model performance was evaluated by comparing the histograms of the automatically estimated characteristic frequencies with those of the manually identified frequencies, showing a good agreement between the two distributions. Furthermore, the Absolute Knee Frequency Error and the Absolute Rain Transition Frequency Error exhibited values close to zero, highlighting the high accuracy of the proposed method. The validity of the proposed model was assessed through a statistical analysis carried out on both a representative sample dataset and the complete database of events acquired at the Tito Scalo and Spino d’Adda ground stations. The consistency of the model was further assessed by comparing the crosswind velocities estimated from the knee frequencies identified by the algorithm in the Ka and Q bands. The analysis showed good agreement between the two estimates, with generally limited errors, confirming the overall coherence of the proposed method. The differences observed between the two analyzed stations are consistent with the performance of the models adopted for the estimation of the knee frequency.The analysis also revealed significant correlations between the characteristic frequencies and the attenuation intensity, especially the mean attenuation, enabling a better characterization of the spectral behavior of the events. Finally, the identified frequencies were used to implement an adaptive filtering procedure aimed at the automatic separation of rain attenuation and tropospheric scintillation components. The visual comparison between the total measured attenuation and the two components obtained from the filtering process showed a coherent reconstruction of the original signal, with an almost complete overlap between the measured signal and the sum of the separated components. This result provides additional confirmation of the correctness of the cutoff frequencies identified by the model and of the physical consistency of the adopted separation procedure. The obtained results demonstrate the validity of the proposed approach and its capability to correctly identify the spectral characteristics of the analyzed events. As future developments, further refinement of the frequency interval selection used for model construction is envisaged, together with the investigation of dedicated strategies for handling particular spectral configurations that do not exhibit the typical behavior observed in most of the analyzed events.| File | Dimensione | Formato | |
|---|---|---|---|
|
2026_07_Malentacchi_Executive_Summary.pdf
accessibile in internet per tutti a partire dal 01/07/2027
Dimensione
903.62 kB
Formato
Adobe PDF
|
903.62 kB | Adobe PDF | Visualizza/Apri |
|
2026_07_Malentacchi_Tesi.pdf
accessibile in internet per tutti a partire dal 01/07/2027
Dimensione
9.14 MB
Formato
Adobe PDF
|
9.14 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/10589/261441