This paper discusses the use of diagnostics based on machine learning (ML) within a flight test context. The paper begins by discussing some of the problems associated with instrumenting a test aircraft and how they could be ameliorated using ML-based diagnostics. We then describe a number of types of supervised ML algorithms which can be used in this context. In addition, key practical aspects of applying these algorithms, such as feature engineering and parameter selection, are also discussed. The paper then outlines a real-world application developed by Curtiss-Wright, called Machine Learning for Advanced System Diagnostics (MLASD). This description includes key challenges that were encountered during the development process and how suitable input features were identified. Real-world results are also presented. Finally, we suggest some further applications of ML techniques, in addition to describing other areas of development.

As Ethernet based networks have become the dominant choice for Flight Test Instrumentation (FTI) network applications, it is also clear that Ethernet based camera integration and applications have yet to become more wide spread for system level design and integration. A significant customer base utilizes either separate video compression systems or even just stand-a-lone gopro cameras for recording purposes in an unsynchronized ways. The use of uncompressed high definition (HD) video from GigE Vision Ethernet cameras for flight test applications is a significant issue in managing the large volumes of data produced by the cameras and forwarding them to any 1000BASE-T(x) switch port without packet loss and significant delays. Of course an easy approach to overcome this issue would be to just increase the network bandwidth from 1000BASE-T(x) to 10GBASE-SR, but most FTI systems just moved to 1000BASE-T(x) in the past years and therefore changing the overall system hardware is cost prohibited. One concern has been the use of compression algorithms to reduce the required video bandwidth, with the negative side effect that the image quality reduces and end-to-end latency increases, which is not acceptable for some applications. Further, it is important that data from cameras is available to a number of different multicast consumers within the FTI network, for example workstations, recorders and telemetry systems. These video data stream also require synchronization so that they can be analyzed in post processing.

In this paper we compute a lower bound, namely the dependence testing (DT) bound, on the maximum achievable rates (expressed in bits/channel use) with military standard shaped-offset quadrature phase shift-keying (SOQPSK-MIL) and aeronautical telemetry SOQPSK (SOQPSK-TG) schemes over additive white Gaussian noise (AWGN) channels under finite code blocklength, probability block error and equiprobable input constraints. The DT bound results for SOQPSK-MIL and SOQPSK-TG are used to lower bound their respective spectral efficiencies (expressed in bits/s/Hz). We simulate a serially concatenated convolutional code (SCCC) using SOQPSK-MIL as the inner code, and show that it performs within 1:1 dB of the SOQPSK-MIL DT bound for various coding rates. The numerical results also demonstrate the performance loss compared to the channel capacity due to the finite blocklength constraint.

Generalized space shift keying (GSSK) is a transmission scheme where only antenna indices are used to send information from the transmitter to a receiver. This paper investigates the best symbol set selection problem in GSSK multiple-input-multiple-output (MIMO) systems when the transmit antennas are correlated. Although multiple antennas can increase data rate and signal quality without increasing the bandwidth, spatial correlations among the antennas highly affect the performance of the system. The idea here is to maximize the inter-symbol Euclidean distance to obtain the best symbol set. Recently such an algorithm has been proposed for the visible light communication (VLC) systems. This paper adopts this VLC algorithm for radio frequency (RF) communication systems. The results show that the proposed symbol set design can provide several dBs of gain in the symbol error rate (SER) performance over randomly selected symbol sets in GSSK systems.

SSDs are now commonplace in all types of computing from consumer laptops to enterprise storage systems. However, most of those SSDs would not survive in environments with extreme temperatures or high shock and vibration such as found in embedded and military systems. The problems in this space are more than just mechanical; they involve all aspects of the design including electrical and even firmware. A combination of all three engineering disciplines is needed to provide a robust ruggedized SSD product.

The modalities of many imaging applications have been advancing toward the MIMO format for improved efficiency. This paper presents an approach to the implementation of the MIMO operating modality for FMCW radar imaging systems with multi-element data-acquisition arrays. The unique characteristics of the software-defined step-frequency FMCW imaging systems enable the simple and precise orthogonalization procedure through direct spectral partitioning. This paper includes the description of the data-acquisition hardware of an eighttransceiver radar imaging system, imaging formation algorithms, MIMO implementation, and results of laboratory experiments.

Many signal processing, data analysis and graphical user interface algorithms are computationally intensive. This paper investigates a method of off-loading some of the calculations to remotely located processors. Inexpensive, commercial off the shelf processors are used to perform operations such as fast Fourier transforms and other numerically intensive algorithms. The data is passed to the processors, and results collected, using conventional network interfaces such as TCP/IP. This allows the processors to be located at any location, and also allows potentially large caches of processors to be shared between multiple applications.

In a previous paper “Merging Multiple Telemetry Files From Widely Separated Sources For Improved Data Integrity” presented at the 2012 ITC\USA conference, a method for synchronizing TM files at the minor frame level was presented. This paper expands on that work by describing a method for synchronizing the files at the minor frame level faster and at the earliest frame possible using an internal counter. This method is also useful if the minor frames fall out of sync due to large dropouts.

Degradation in visibility is often introduced to images captured in poor weather conditions, such as fog or haze. In this paper, a fusion-based transmission estimation method is introduced to adaptively combine two different transmission models. Specifically, the new fusion weighting scheme and the atmospheric light computed from the Gaussian-based dark channel method improves the estimation of the locations of the light sources. To reduce the flickering effect introduced during the process of frame-based dehazing, a flicker-free module is formulated to alleviate the impacts. The system assessments show this approach is capable of superior defogging and dehazing performance, compared to the state-of-the-art methods, both quantitatively and qualitatively

This document will provide a detailed description of the original design behind our device, device casing, and iOS application. It will cover process of assembly, as well as failure analysis and future directions for the project.