Crazy Propeller 2 Carrier-Phase DGPS application ?
Greg P
Posts: 58
Chip,
Is it crazy to imagine the Propeller 2 being capable of performing 5 to 10 Mhz multiplication of the 2 or 3 bit GPS IF-freq ADC's output with both a cos and sine function ? This would be direct IF to baseband mixing of the raw GPS signal. The result of these multiplications would then require lower frequency (say 20khz) multiplications by additional cos and sin functions to correct for satellite-dependent doppler frequency shifts. The neat trick in using the algorithm described by Petovello in the article below is that the high-rate 5 to 10 Mhz math need only be applied once. The lower frequency 20 kHz math addresses a linearly accumulating phase shift due to doppler error. Once this error is eliminated (and of course after c/a code despreading), the resulting I and Q values yield the "carrier phase", the difference in phase between the satellite's carrier and the local receiver oscillator. Note that the small frequent frequency errors are fed to a digital filter which then adjusts at a 20khz rate the sin/cos values used by the doppler elimination math, so that high-quality frequency tracking is maintained.
If single cycle multiplication, 32-bit sin functions, and 160 MIPS/cog is actually possible with the Propeller 2, that's 16 cycles available at 10Mhz sampling. Could one cog do the high bandwidth multiplication then share its data stream result with other cogs ?
I know it sounds crazy to want to do this with a Propeller when other integrated solutions already exist, but I'm concerned about multiplying cost when two or even three GPS antennas and their respective receivers are integrated together for carrier-phase based DGPS purposes. If the Propeller 2 will indeed support high speed serial links between multiple chips, it seems at least possible to combine these chips into an GPS engine capable of handling multiple GPS antennas without the need for individual receivers.
REFERENCE:
Centimeter-level positioning using an efficient new baseband mixing
and despreading method for software GNSS receivers
EURASIP Journal on Advances in Signal Processing
Volume 2008 , (January 2008), Article No. 4
A copy of the PDF article may be found here:
www.hindawi.com/journals/asp/2008/194276.pdf
Is it crazy to imagine the Propeller 2 being capable of performing 5 to 10 Mhz multiplication of the 2 or 3 bit GPS IF-freq ADC's output with both a cos and sine function ? This would be direct IF to baseband mixing of the raw GPS signal. The result of these multiplications would then require lower frequency (say 20khz) multiplications by additional cos and sin functions to correct for satellite-dependent doppler frequency shifts. The neat trick in using the algorithm described by Petovello in the article below is that the high-rate 5 to 10 Mhz math need only be applied once. The lower frequency 20 kHz math addresses a linearly accumulating phase shift due to doppler error. Once this error is eliminated (and of course after c/a code despreading), the resulting I and Q values yield the "carrier phase", the difference in phase between the satellite's carrier and the local receiver oscillator. Note that the small frequent frequency errors are fed to a digital filter which then adjusts at a 20khz rate the sin/cos values used by the doppler elimination math, so that high-quality frequency tracking is maintained.
If single cycle multiplication, 32-bit sin functions, and 160 MIPS/cog is actually possible with the Propeller 2, that's 16 cycles available at 10Mhz sampling. Could one cog do the high bandwidth multiplication then share its data stream result with other cogs ?
I know it sounds crazy to want to do this with a Propeller when other integrated solutions already exist, but I'm concerned about multiplying cost when two or even three GPS antennas and their respective receivers are integrated together for carrier-phase based DGPS purposes. If the Propeller 2 will indeed support high speed serial links between multiple chips, it seems at least possible to combine these chips into an GPS engine capable of handling multiple GPS antennas without the need for individual receivers.
REFERENCE:
Centimeter-level positioning using an efficient new baseband mixing
and despreading method for software GNSS receivers
EURASIP Journal on Advances in Signal Processing
Volume 2008 , (January 2008), Article No. 4
A copy of the PDF article may be found here:
www.hindawi.com/journals/asp/2008/194276.pdf
Comments
Prop 1 is capable for 2-3 bit multiplication at 10 MHz with 4 COGs if you can distribute/collect the data so fast·to/from them.
Cheers,
Istvan