The 7 th National Radar Seminar and The 2 nd Inter national Confer ence on Radar , Antenna, Micr owave, Electr onics and Telecommunications (ICRAMET) Sur abaya – Indonesia | 27 – 28 Mar ch 2013 Pa g e | 61 ICRAMET SAR Raw Data Compression based on Compressive Sensing Rahmat Arief 1 , Musyarofah 1 , Kalamullah Ramli 2 1 Remote Sensing Technology and Data Center, Indonesian National Institute of Aeronautics and Space (LAPAN) Jl. Lapan No 70, Jakarta Timur 13710 2 Departement of Electronic Engineering, University of Indonesia Email : rahmat.arief @lapan.go.id Abstract—Recently, the use of SAR image data is rapidly growing due to the ability of SAR sensor to generate high resolution and wide swath data, to operate day and night, penetrate cloud, and provide complementary data for optical sensor. However, SAR sensor system needs large data volume and high electricity power to obtain a good quality of SAR image data. Compressive sampling, as a new method, gives solution to overcome those problems. With this method, the SAR image of sparse targets can be recovered by solving the convex optimization problem with a very few of SAR echo samples under Nyquist/Shannon theorem required. Therefore, it can reduce data volume, and electricity power as well. Analysis was conducted by comparing the image quality values of raw SAR data, using compressive sampling method and the conventional one. Point target in raw SAR data was particularly selected to give an obvious illustration of the image quality parameters. Point target of simulation and Radarsat-1 raw data were compared by simulation to get their parameter values. From the simulation, it showed that raw SAR data using compressive sampling method could reduce an amount of sidelobes, and so, spatial resolution of SAR image data would be increasing. Keywords-SAR, raw data, compressive sampling method, point target, PSLR, ISLR. I. INTRODUCTION Development of radar technology leads the increase of Synthetic Aperture Radar (SAR) image data use. SAR image data has several advantages if compared to optical image data. SAR sensor is allowed to operate day and night, penetrate cloud, and so the data resulted is free cloud and the number could be more than optical sensor. Moreover, SAR image data gives different information about target, therefore it is able to provide an additional information as a complementary data for optical image. Nowadays, SAR sensor technology is growing rapidly, it has an ability to generate high resolution and wide swath image data. One of the main challenges in the current SAR imaging based on Shannon / Nyquist sampling method, which requires a high rate Analog Digital Converter (ADC) on the limited storage resources on-board, bandwidth and computational processing. Thus, large volumes of SAR data must be directly stored on board and then transmitted to the ground in some compressed form for further processing. These problems can be solved with compressive sensing (CS) method. With this method, the SAR image of sparse targets can be recovered by solving the convex optimization problem with a very few of SAR echo samples under Nyquist/Shannon theorem required [1] [2]. Unlike conventional compression method in this context, in compressed sensing (CS) we do not measure the K non zero element directly in during signal/image acquisition. Rather, we measure and encode M < N of the signal. Recovery of the signal from the measurements is in general ill-posed. However, the CS theory declare that is possible to recover the K largest from a set of M measurements. CS theory suggested to move the computational load of the onboard side to the ground receiver side. On the onboard side, the signal obtained with a sampling rate less than the Shannon-Nyquist rate. Then, at the ground receiver, the measurement is used for the reconstruction of the original signal. Several studies CS applications for SAR data have been proposed. The results of experiment and simulation show the effectiveness of this theory for SAR data compression. In this paper, we study the use of compressive sensing methods to reduce SAR raw data. We mainly focus on random sampling on the receiving part of the SAR system and discussion of optimal recovery methods and analysis of image quality values of SAR image. To give an obvious illustration, a sample scene of simulated point targets and Radatsat-1 raw data was particularly selected, and then determined its image quality values. From that comparation, the reduction could be analyzed.Further, we offer opinions on the potential impact of CS insights on future directions of application of CS in SAR system. II. COMPRESSIVE SENSING BASIC The CS enables the reconstruction of sparse or compressible signals from a small set of non adaptive, linear measurements. The main interest in compressed sensing research is by solving of the inverse problem of a vector signal ( ݐ)= ∑ ݔ ௜  ௜ ( ݐ), ே ௜ୀଵ ∈ ே that it is sparse in some basis :  = Ψ ݔwhere Ψ∈ ெ௫ே is a matrix whose columns are the