There is a lot to answer in this.
So, let's talk about light how sunlight affects the result. Sunlight has 940nm light in its spectrum. And although there is less of it - because 940nm light is absorbed by water vapor in the upper atmosphere it's still significant. So the signal to noise ratio is the amount of signal the sensor can see above the sunlight's 'noise'.
But perhaps you meant noise on the power rails. That's pretty much handled by the sensor. We invented it for the cell-phone market and put a lot of power conditioning in due to the expected battery power drop. So, it's not really an issue. (But don't be *** about it. Put those recommended caps in there. )
Now I2C noise is the real killer. As a bit is clocked as the CLK pin drops, any glitch during this decent causes an extra bit to clock. And once the host and the sensor get out of sync, you end up with the famous 'line-stuck-low' issue.
So, choosing your I2C configuration is really important. Keep your wires short. Keep them thick and good insulation helps. Sometimes twisted pair works better. And google how to choose a pull-up based on the capacitance of your system. (I'm software - I've never really understood the formula that's used to do this.)
It's not often that you need a checksum on the I2C as any minor glitch will just hang the bus.
Good luck,
- john
