Constrained RS coding for Low Peak to Average Power Ratio in FBMC -- OQAM Systems

Multi-carrier modulation techniques have now become a standard in many communication protocols. Filter bank based multi-carrier (FBMC) generation techniques have been discussed in the literature as a means for overcoming the shortcomings of IFFT/FFT based OFDM system. The Peak to Average Power Ratio (PAPR) is a problem faced by all multi-carrier techniques. This paper discusses the methods for reducing PAPR in a FBMC system while maintaining acceptable Bit Error Rate (BER). A new PAPR minimizing scheme called Constrained Reed Solomon (CRS) coding is proposed. The hybrid techniques using coding and companding are tested for different channel models and is found to yield promising results.

IR Motion Deblurring

Camera gimbal systems are important in various air or water borne systems for applications such as navigation, target tracking, security and surveillance. A higher steering rate (rotation angle per second) of gimbal is preferable for real-time applications since a given field-of-view (FOV) can be revisited within a short period of time. However, due to relative motion between the gimbal and scene during the exposure time, the captured video frames can suffer from motion blur. Since most of the post-capture applications require blurfree images, motion deblurring in real-time is an important need. Even though there exist blind deblurring methods which aim to retrieve latent images from blurry inputs, they are constrained by very high-dimensional optimization thus incurring large execution times. On the other hand, deep learning methods for motion deblurring, though fast, do not generalize satisfactorily to different domains (e.g., air, water, etc). In this work, we address the problem of real-time motion deblurring in infrared (IR) images captured by a gimbal-based system. We reveal how a priori knowledge of the blur-kernel can be used in conjunction with non-blind deblurring methods to achieve real-time performance. Importantly, our mathematical model can be leveraged to create large-scale datasets with realistic gimbal motion blur. Such datasets which are a rarity can be a valuable asset for contemporary deep learning methods. We show that, in comparison to the state-of-the-art techniques in deblurring, our method is better suited for practical gimbal-based imaging systems.