canoeguy1 :
To get more technical, an indirect path may be attenuated by 20 dB vs a direct path, but if the light is strong enough to provide a decent SNR, it's not an issue.
20dB attenuation? I think the figure would be over 100dB: your LED's beam will expand to 1m if not more, which already makes it 60+dB weaker, the paint will reflect around 50% of it, only a tiny fraction of that diffused reflection will be going towards the sensor unless the sensor is conveniently located in the normal reflection angle and will likely cover only a tiny percentage of the sensor's field-of-view with the rest contributing to noise input.
Indirect may work fine on the downstream if the flood light is the one getting modulated for that. For the upstream though, you would likely need to have a sensor network to cover blind spots and divide rooms into multiple sensor zones to reduce the amount of optical noise per zone.
Can it work? Sure - this is not the first time free-air optical transmissions have been achieved, albeit in fixed point-to-point arrangements. Would it be a practical WiFi replacement? I don't think so. With LiFi, you would need to replace most of your lightbulbs with $?? LiFi lightbulbs, can only use your LiFi mobile devices where LiFi bulbs are installed and need to keep your device's LiFi transceiver location in mind to avoid blocking it. With WiFi, you install a WAP/router anywhere convenient and can use your WiFi devices anywhere within range in any orientation you want.
Also, WiFi passes through clothes so you can receive your Facebook/Google/Skype/VoIP call/whatever notifications over WiFi while your device is in your pocket/purse/binder/whatever. Can't do that with LiFi no matter how good it may become.