<p> One of the important objectives of a Radar Warning Receiver (RWR) aboard a tactical aircraft is to evaluate the level of threat posed by hostile radars in an extremely complex Electronic Warfare (EW) environment in reliable, robust and timely manner. For the RWR objective to be achieved, it passively collects electromagnetic signals emitted from potentially hostile radars. One class of such radar systems is the Multi-Function Radar (MFR) which presents a serious threat from the stand point of a RWR. MFRs perform multiple functions simultaneously employing complex hierarchical signal architecture. The purpose of this paper is to uncover the evolution of the operational mode (radar function) from the view point of a target carrying the RWR when provided with noisy observations and some prior knowledge about how the observed radar functions. The RWR estimates the radar's threat which is directly dependant on its current mode of operation. This paper presents a grid filter approach to estimate operational mode probabilities accurately with the aid of pre-trained Observable Operator Models (OOMs) and Hidden Markov Models (HMMs). Subsequently, the current mode of operation of a radar is estimated in the maximum a posteriori (MAP) sense. Practicality of this novel approach is tested for an EW scenario in this paper by means of a hypothetical MFR example. Finally, we conclude that the OOM-based grid filter tracks the mode of operation of a MFR more accurately than the corresponding HMM-based grid filter. </p> / Thesis / Master of Applied Science (MASc)

<p>Radar warning systems are life saving equipment in modern fighter aircraft. It is therefore vital that the system can tell the difference between a threat genuine frequency) and a false signal (spurious frequency). </p><p>This thesis presents a model aimed at predicting the frequencies and other parameters in the RF converter of the radar warning system. The components of the RF converter have been studied, measured, and modelled. The modelling tool has been the Simulink toolbox for Matlab. </p><p>Extreme accuracy has been sacrificed in order to make the model easy to use for the working engineer. Instead, this model presents a rough estimate of some of the most important properties of the radar warning system with just a few data sheet figures as input.</p><p>The simulation results are satisfactory as a whole. Simulink is the limiting factor in the implementation of the model. Significantly improved results can probably be obtained by working in another software environment.</p>

Radar warning systems are life saving equipment in modern fighter aircraft. It is therefore vital that the system can tell the difference between a threat genuine frequency) and a false signal (spurious frequency). This thesis presents a model aimed at predicting the frequencies and other parameters in the RF converter of the radar warning system. The components of the RF converter have been studied, measured, and modelled. The modelling tool has been the Simulink toolbox for Matlab. Extreme accuracy has been sacrificed in order to make the model easy to use for the working engineer. Instead, this model presents a rough estimate of some of the most important properties of the radar warning system with just a few data sheet figures as input. The simulation results are satisfactory as a whole. Simulink is the limiting factor in the implementation of the model. Significantly improved results can probably be obtained by working in another software environment.

The radio frequency spectrum is becoming increasingly crowded and research efforts are being made both from the side of communication and from radar to allow for sharing of the radio frequency spectrum. In this thesis, suitable methods for classifying incoming signals as either communication signals or radar signals using machine learning are evaluated, with the purpose of filtering communication signals in radar warning receiver systems. To this end, a dataset of simulated communication and radar signals is generated for evaluation. The methods are evaluated in terms of both accuracy and computational complexity since both of these aspects are critical in a radar warning receiver setting. The results show that a deep learning model can be designed to outperform expert feature-based models in terms of accuracy, as has previously been confirmed in other fields. In terms of computational complexity, however, they are vastly outperformed by a model based on ensemble decision trees. As such, a deep learning model may be too complex for the task of filtering communication signals from radar signals in a radar warning receiver setting. The classification accuracy needs to be weighed against the model size and classification time. Future work should focus on optimizing the feature extraction implementation for a more fair classification time comparison, as well as evaluating the models on recorded data. / Radiospektrumet blir alltmer belastat och forskningsinsatser görs inom både kommunikation och radar för att tillåta delning av spektrumet. I denna rapport utvärderas lämpliga metoder för att klassificera inkommande signaler som antingen kommunikation eller radar med hjälp av maskininlärning, med syftet att filtrera ut kommunikationssignaler i radarvarnare. För detta ändamål genereras ett dataset med simulerade kommunikations- och radarsignaler för att jämföra modellerna. Metoderna utvärderas med avseende på både precision och beräkningskomplexitet, eftersom att båda aspekterna är kritiska egenskaper i en radarvarnare. Resultaten visar att en djupinlärningsmodell kan utformas för att överträffa modeller baserade på expertdesignade särdrag med avseende på träffsäkerhet, såsom tidigare visats inom andra områden. Avseende beräkningskomplexitet, är däremot modellen baserad på en ensemble av beslutsträd överlägsen. Detta innebär möjligen att en djupinlärningsmodell är allt för komplex för syftet att filtrera bort kommunikationssignaler från radarsignaler i en radarvarnare. Modellens träffsäkerhet bör vägas mot dess storlek och tiden för klassificering. Framtida arbete bör inriktas på att optimera beräkningen av särdragen för en mer rättvis jämförelse av tiden som krävs för klassificering, samt att utvärdera modellerna på inspelad data.

Approved for public release, distribution is unlimited / Presently, the required data file to be loaded onto the Radar Warning Receiver (RWR) onboard the F-16 aircraft is done manually by the aircraft technicians, two to three hours prior to the actual flight time. This process should be automated. As such there is a need to look into the use of wireless transmission technology to complement or replace the manual method of loading the critical data file from the command station onto every F-16 aircraft. The present wireless technology is relatively mature and stable. In this thesis, the feasibility of incorporating and adapting this technology for use in the flight line environment is examined. The propagation effect in wireless transmission is also studied and recommendations proposed with regards to the installation of wireless facilities in the flight line. In addition, the EDNA, a portable maintenance aid that comes with the F-16 aircraft for loading the data file, has to be upgraded. Hence, a system feasibility study is carried out to adapt or upgrade the present equipment to wireless transmission capability. / Major, Republic of Singapore Air Force

Radar signals are used for estimating location, speed and direction of an object. Some radars emit pulses, while others emit a continuous wave. Both types of radars emit signals according to some pattern; a pulse radar, for example, emits pulses with a specific time interval between pulses. This time interval may either be stable, change linearly, or follow some other pattern. The interval between two emitted pulses is often referred to as the pulse repetition interval (PRI), and the pattern that defines the PRI is often referred to as the modulation. Classifying which PRI modulation is used in a radar signal is a crucial component for the task of identifying who is emitting the signal. Incorrectly classifying the used modulation can lead to an incorrect guess of the identity of the agent emitting the signal, and can as a consequence be fatal. This work investigates how a long short-term memory (LSTM) neural network performs compared to a state of the art feature extraction neural network (FE-MLP) approach for the task of classifying PRI modulation. The results indicate that the proposed LSTM model performs consistently better than the FE-MLP approach across all tested noise levels. The downside of the proposed LSTM model is that it is significantly more complex than the FE-MLP approach. Future work could investigate if the LSTM model is too complex to use in a real world setting where computing power may be limited. Additionally, the LSTM model can, in a trivial manner, be modified to support more modulations than those tested in this work. Hence, future work could also evaluate how the proposed LSTM model performs when support for more modulations is added. / Radarsignaler används för att uppskatta plats, hastighet och riktning av objekt. Vissa radarer sänder ut signaler i form av pulser, medan andra sänder ut en kontinuerlig våg. Båda typer av radarer avger signaler enligt ett visst mönster, till exempel avger en pulsradar pulser med ett specifikt tidsintervall mellan pulserna. Detta tidsintervall kan antingen vara konstant, förändras linjärt, eller följa ett annat mönster. Intervallet mellan två pulser benämns ofta pulsrepetitionsintervall (PRI), och mönstret som definierar PRIn benämns ofta modulering. Att klassificera vilken PRI-modulering som används i en radarsignal är en viktig del i processen att identifiera vem som skickade ut signalen. Felaktig klassificering av den använda moduleringen kan leda till en felaktig gissning av identiteten av agenten som skickade ut signalen, vilket kan leda till ett dödligt utfall. Detta arbete undersöker hur väl det framtagna neurala nätverket som består av ett långt korttidsminne (LSTM) kan klassificera PRI-modulering i förhållande till en modern modell som använder särskilt utvalda beräknade särdrag från data och klassificerar dessa särdrag med ett neuralt nätverk. Resultaten indikerar att LSTM-modellen konsekvent klassificerar med högre träffsäkerhet än modellen som använder särdrag, vilket gäller för alla testade brusnivåer. Nackdelen med LSTM-modellen är att den är mer komplex än modellen som använder särdrag. Framtida arbete kan undersöka om LSTM-modellen är för komplex för att använda i ett verkligt scenario där beräkningskraften kan vara begränsad. Dessutom skulle framtida arbete kunna utvärdera hur väl LSTM-modellen kan klassificera PRI-moduleringar när stöd för fler moduleringar än de som testats i detta arbete läggs till, detta då stöd för ytterligare PRI-moduleringar kan läggas till i LSTM-modellen på ett trivialt sätt.