Loading...
Please wait, while we are loading the content...
Similar Documents
Using GPU VSIPL & CUDA to Accelerate RF Clutter Simulation
| Content Provider | Semantic Scholar |
|---|---|
| Author | Campbell, Dan McCans, Mark Davis, Mike Brinkmann, Mike |
| Copyright Year | 2011 |
| Abstract | This paper describes a flexible simulator for background Radio Frequency clutter developed at the Georgia Tech Research Institute, and how this simulation was accelerated with the use of nVidia GPUs using GPU VSIPL. The paper describes the mathematical basis for the simulation and how it can be used to simulate RF environments and scenarios; introduces the VSIPL API; describes the porting and validation process; highlights challenges raised by the conversion from double to single precision, and how they were met; and describes the techniques used to obtain improved execution speed, achieving 70x improvement over the original simulation. I. RF CLUTTER SIMULATION The RF clutter simulation is a portion of a testbed to evaluate the performance of radar components and radar algorithms. In order to present realistic simulated radar return data, realistic simulated background clutter must be generated. GTRI maintains a flexible radar environment simulator developed in MATLAB, which contains a module to create this simulated data. A radar system senses its environment by transmitting radio frequency (RF) energy and observing the echoes from objects in the surveillance area. In many scenarios of interest, the performance of such a system is limited by the strong return from the ground, which can mask smaller targets. In order to understand the impact of this so called radar clutter, and evaluate methods to suppress it, high fidelity simulations are required. While it is straightforward to simulate radar returns from discrete targets, simulation of the return from every point on the ground that is visible to a radar is far more challenging. In many cases radar systems are physically capable of resolving patches on the ground that are less than a meter on each side. A high fidelity clutter simulation must simulate scatterers finer than this spacing over points on the earth that can extend for thousands of kilometers. This approach is necessary since the contribution due to each resolvable clutter patch will have distinct angle and Doppler properties. These properties must be faithfully reproduced since it this structure that allows a radar signal processor to suppress clutter. The complexity of radar clutter simulations is due precisely to this fact: each point on the ground must be treated independently in order to model the salient features of the corresponding radar data. In contrast to the computational complexity, the concept behind the radar clutter simulation is relatively simple. A radar is essentially only capable of measuring range. Because all targets at a particular range will be observed by the radar at the same time, the clutter simulation first divides the ground into range rings. Due to the distributed nature of clutter, each range ring may be processed independently. This range ring is then further sub-divided into a number of clutter patches. Each clutter patch may be described according to its range, azimuth, and elevation relative to the platform. In order to determine the contribution of a clutter patch, a model of the radar system is then employed. The radar range equation maps these (range, azimuth, elevation) coordinates along with the radar system parameters to determine how much energy each patch contributes. This process is repeated for each clutter patch in a given range ring, for each range ring, and ultimately for each of several to many pulses. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://saahpc.ncsa.illinois.edu/10/presentations/day2/session4/presentation_Campbell.pdf |
| Alternate Webpage(s) | http://saahpc.ncsa.illinois.edu/10/papers/paper_5.pdf |
| Alternate Webpage(s) | https://www.ll.mit.edu//HPEC/agendas/proc10/Day2/F3_1030_Campbell_RF_Clutter_presentation.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |