Accelerating a Cloud-Based Software GNSS Receiver

TL;DR

This paper explores methods to accelerate a cloud-based software GNSS receiver by leveraging multi-core CPUs and GPGPUs, aiming to minimize processing time per request and maximize overall system throughput.

Contribution

It introduces specific techniques for optimizing performance and resource management in cloud-based GNSS processing, including effective resource allocation and throughput measurement.

Findings

Speedups achieved through multi-core CPU utilization

GPGPU acceleration techniques improved processing times

Resource control strategies increased system throughput

Abstract

In this paper we discuss ways to reduce the execution time of a software Global Navigation Satellite System (GNSS) receiver that is meant for offline operation in a cloud environment. Client devices record satellite signals they receive, and send them to the cloud, to be processed by this software. The goal of this project is for each client request to be processed as fast as possible, but also to increase total system throughput by making sure as many requests as possible are processed within a unit of time. The characteristics of our application provided both opportunities and challenges for increasing performance. We describe the speedups we obtained by enabling the software to exploit multi-core CPUs and GPGPUs. We mention which techniques worked for us and which did not. To increase throughput, we describe how we control the resources allocated to each invocation of the software to process a client request, such that multiple copies of the application can run at the same time. We use the notion of effective running time to measure the system's throughput when running multiple instances at the same time, and show how we can determine when the system's computing resources have been saturated.