The frequency-shift (FRESH) filter is a structure which exploits the spectral correlation of cyclostationary signals for removing interference and noise from a wanted signal. As most digital communication signals are cyclostationary, FRESH filtering offers certain advantages for interference rejection in a communications receiver. This thesis explores the operation and application of FRESH filters in practical interference scenarios. The theoretical background to cyclostationarity is clarified with graphical interpretations of what cyclostationarity is, and how a FRESH filter exploits it to remove interference. The effects of implementation in a sampled system are investigated, in filters which use baud rate related cyclostationarity, leading to efficiency improvements. The effects of varying the wanted signal pulse shape to enhance the cyclostationarity available to the FRESH filter are also investigated. A consistent approach to the interpretation of the FRESH filter's operation is used throughout, while evaluating the performance in a wide range of realistic channel conditions. VLF radio communication is proposed as one area where interference conditions are particularly suitable for the use of FRESH filtering. In cases of severe adjacent channel interference it is found that a FRESH filter can almost completely remove the interferer. The effects of its use with an impulse rejection technique are also investigated. Finally, blind adaptation of FRESH filters through exploitation of carrier related cyclostationarity is investigated. It is found that one existing method loses the advantage of FRESH filtering over time invariant linear filtering. An improvement is proposed to the latter which restores its performance to that of a trained FRESH filter, and also reveals that carrier related cyclostationarity can be exploited, in some cases, by a simpler method. J.

Swedish railway network is old, it is in need for maintenance and there is an ever-increasing volume oftraffic on todays railroads. Special trackwork component has taken increased abuse, leading to a stop intrain traffic and many delays occur, especially during the winter season. The most sensitive parts of therailway network are turnout frogs and they require the most maintenance and cause delays. Therefore, TheSwedish Transport Administration is planning to investigate and introduce a new type av railwayswitches, called flange bearing technology. Flange bearing frogs and diamonds are already in use in USand there are already benefits in both reduced maintenance and increased life expectancy and possiblyreduced noise level.Therefore, this bachelor thesis is investigating flange bearing technology with an accent on sound levelmeasurement. It is studied how flange bearing frogs works and how they differ from conventional treadbearing frogs. In addition, there is a field study with sound level measurement and interviews withemployees at The Swedish Transport Administration. Furthermore, it is investigated how flange bearingtechnology is used in US. This bachelor thesis is limited to a field research with sound level measurement,interviews, and a theoretical description of railway switches. Field study is located at a flange bearingfrog, station Globen-Gullmarsplan in Stockholm. This flange bearing frog is used as a connection betweenthe tram railway and the metro railway.Flange bearing technology has a simpler design compared to conventional tread bearing frogs. This simpledesign can provide benefits by reducing vertical dynamic loads. Flange bearing means that unsupportedflangeway gap is eliminated from the running surface of the frog. For that reason, there is no speedreduction on the main track. Conventional tread bearing frogs have unsupported flangeway gap thatwheels must cross.Field study at Globen-Gullmarsplan was carried out with the help of a sound level meter and it wasperformed eight measurements at both flange bearing frog and conventional tread bearing frog. This studyshows that flange bearing frogs causes less noise than conventional tread bearing frog. Interviews showsthat The Swedish transport Administration is working to reduce noise interference from rail traffic andthat flange bearing technology can be a good solution for achieving lower noise level. However, moreresearch is required about maintenance, comfort, cost, and life expectancy of flange bearing technology.

Within this thesis, we investigate the possibilities of physical layer secrecy for two special system models. In detail, we study beamforming and protection strategies in the Multiple-Input Single-Output (MISO) Gaussian Wiretap Channel (WTC) and the Gaussian two-hop relay WTC with multiple antennas at transmitter and receiver. In both system models, we examine the influence of partial Channel State Information (CSI) on the link to the eavesdropper and compare the achievable secrecy rates with the case of full CSI. We show for the MISO WTC that in the fast fading scenario the Beamforming Vector (BV) can be optimized such that the ergodic secrecy rate is maximized with regard to the degree of channel knowledge. Further we show that the ergodic secrecy rate can be significantly increased by usage of Artificial Noise (AN), if applied in a smart way. This means that the degree of channel knowledge on the link to the eavesdropper influences the portion of power that is spent for AN at the transmitter as well as the direction, in which the AN signal is sent. In addition, we apply the same beamforming and protection strategies to the slow fading scenario and find that these techniques also reduce the secrecy outage probability. For the two-hop relay WTC, we introduce Information Leakage Neutralization (IN) as a new protection strategy. If applied to a system model, where the transmitter has full CSI, the instantaneous secrecy rate performs almost as well as the instantaneous capacity of the peaceful system without an eavesdropper. The IN protected scheme outperforms the AN protected approach and performs much better than any beamforming scheme without additional protection mechanism. Another positive aspect of the IN protected scheme in the case of full CSI is that conventional channel codes can be applied instead of wiretap codes. For the case of partial CSI, where the transmitter has only an outdated estimate on the channel between relay and the eavesdropper, we show that the IN protected scheme can also be applied. Here, it strongly depends on the channel realizations and the delay of the estimate, whether the IN or the AN protection scheme should be applied. / In dieser Arbeit wird das Leistungsvermögen der Sicherheit auf der physikalischen Schicht anhand von zwei speziellen Systemmodellen untersucht. Im Detail werden Beamforming- und Absicherungsstrategien im gaußschen Multiple-Input Single-Output (MISO) Wiretap Channel (WTC) und dem gaußschen Two-hop Relay WTC mit mehreren Antennen am Sender und Empfänger studiert. In beiden Systemmodellen wird der Einfluss von partieller Kanalkenntnis zum Abhörer betrachtet und die so erreichbaren Sicherheitsraten mit denen verglichen, die bei voller Kanalkenntnis erreichbar sind. Für den MISO WTC kann gezeigt werden, dass für Kanäle mit schnellem Schwund der Beamforming-Vektor in Hinblick auf die ergodische Sicherheitsrate unter Berücksichtigung des Grades der Kanalkenntnis optimiert werden kann. Zudem kann durch die intelligente Verwendung von künstlichem Rauschen (Artificial Noise, AN) die ergodische Sicherheitsrate signifikant erhöht werden. Hierbei nimmt der Grad der Kanalkenntnis direkt Einfluss auf die Aufteilung der Leistung zwischen Daten- und AN-Signal am Sender sowie auch auf die Richtung, in der das AN-Signal gesendet wird. Zudem kann gezeigt werden, dass dieselben Beamforming- und Absicherungsstrategien ebenfalls die Sicherheitsausfallwahrscheinlichkeit für Kanäle mit langsamem Schwund minimieren. Im gaußschen Two-hop Relay WTC wird Information Leakage Neutralization (IN) als neuartige Absicherungsstrategie eingeführt. Diese Absicherungsstrategie erreicht nahezu dieselben instantanen Raten wie ein friedvolles System ohne Abhörer, wenn es bei voller Kanalkenntnis am Sender eingesetzt wird. Weiterhin sind durch die IN-Absicherungsstrategie höhere Raten erreichbar als durch den Einsatz von AN. Zusätzlich kann im Fall von voller Kanalkenntnis auf den Einsatz von Wiretap-Codes verzichtet werden. Auch im Fall partieller Kanalkenntnis, wo der Sender nur eine veraltete Schätzung des Kanals zwischen Relay und Abhörer besitzt, kann gezeigt werden, dass die IN-Absicherungsstrategie angewendet werden kann. Hierbei hängt es jedoch stark von den Kanalrealisierungen und dem Alter der Kanalschätzung ab, ob die IN- oder die AN-Absicherungsstrategie bessere Ergebnisse bringt und daher angewandt werden sollte.