On the energy and computational cost of message authentication schemes for GNSS

To produce a position, velocity, and time estimate, modern Global Navigation Satellite System (GNSS) receivers expend energy on two very different kinds of computation: one being high-rate, low resolution, and the other being low-rate, high-resolution. The first kind is largely comprised of integer multiply, or multiply-and-accumulate operations, mostly applied to simple digital signal processing operations, such as frequency-translation, digital filtering, and correlation. These operations generally run at a rate commensurate with the receiver's front-end analog-to-digital converters, which will run at rates of some tens of megahertz, but may only require a few bits of integer resolution. The second kind mostly consists of baseband processing, such as detection, estimation, and tracking, orbit propagation, and position computation. These operations will generally require floating point arithmetic and will run at rates of some tens of Hertz to some kilohertz.