How Lasers Could Transform Wireless Communication

Laying fibre optic cables under city streets is a nightmare for pedestrians and traffic. In a dense Indian city, permits, trenches, and road damage add up to months of chaos. Radio skips the digging but runs into spectrum, which is licensed, auctioned and released on the regulator's schedule. Free space optical communication takes a third route: put a laser on one rooftop and a receiver on another, line the two up, and the link is established. There is nothing to dig up and no licence to wait for, and the crew would have finished by lunch.

Field demonstrations have pushed a single channel past 800 gigabits per second, and others have carried 13 terabits per second across ten kilometres of open air. One terabit per second is roughly 125,000 HD video streams running simultaneously. Imagine data moving through the air like birds. So why is there not one of these bolted to every tall building?

Research led by Dr. Kamal Singh, Assistant Professor in the Department of Electrical Engineering, in collaboration with recent Ph.D. graduate Dr. Himani Verma, now quantifies how much of this performance loss is unavoidable. They have derived an exact expression for the maximum efficiency with which such a link can use its bandwidth when the transmitter adapts to changing conditions in real time.

Wi-Fi can get attenuated through walls, and calls drop in crowded rooms. But for a laser travelling through air, the culprit is temperature. Air at different temperatures bends light by slightly different amounts, so a beam passing over a hot car park is not travelling through the same medium as the beam a metre above it. On a still morning, the light goes more or less straight. On a hot afternoon, it wanders, spreads and flickers, and that flicker arrives at the receiver as bit errors. Engineers describe its statistics with the gamma-gamma distribution, which survives in the literature mostly because it fits the measurements.

The second problem is not optical at all. The mounts holding the transmitter and receiver are not perfectly rigid, so wind pushes them, metal expands as the day heats up, and a tall building sways slightly and continuously. Align a beam precisely at nine in the morning, and by two in the afternoon, it has drifted. Engineers call the leftover misalignment jitter, and there is always some.

Dr. Singh and Dr. Verma focused specifically on coherent free-space optical communication (FSO) systems. Most commercial equipment encodes data by varying the laser's brightness, a cheap and forgiving approach. Coherent systems modulate amplitude and phase together, yielding substantial sensitivity improvements, but the catch is that a receiver sensitive enough to read phase is also sensitive to everything else. Turbulence and jitter included. When both turn up together, the signal-to-noise ratio drops, and throughput follows.

Adaptive transmission is the other half of the arrangement. Rather than sending at a fixed rate and hoping, the system watches the channel and changes its behaviour, pushing hard when conditions are strong and pulling back when they deteriorate. The optimal way to spread power across those fluctuations follows a principle called water filling, which concentrates resources where the channel is strongest and leaves the weak moments largely alone. It is less like a firehose than like irrigation.

The first of two counterintuitive results appears at a high signal-to-noise ratio when jitter is severe. The natural assumption is that stronger turbulence always hurts, and most of the time it does, but turbulence also makes the beam spread as it propagates, and a wider beam is a more forgiving target. If the mount is wobbling by a given amount, a fat beam loses less of itself off the receiver's edge. One impairment partly offsets the other, improving spectral efficiency by roughly 2.8 bits per second per hertz even as the total received power falls.

The second result is stranger. At a low signal-to-noise ratio, where power per unit bandwidth is small because eye safety regulations cap what you are allowed to transmit, or because a fixed budget has been carved up across many optical channels at once, stronger turbulence links better. Not less bad. Better. A more turbulent channel throws up more frequent episodes of unusually high gain, brief windows in which the atmosphere happens to focus the beam rather than scatter it, and an adaptive transmitter can wait for those windows and pour power into them. What it gains there outweighs what it gives up in between.

What the paper delivers is a ceiling, and an exact one, which is rarer than it sounds for this class of problem. Most analyses in the field end in bounds and approximations. A ceiling is not a receiver design, though, and it is not a working link on a roof in Noida in December. The result sits atop a statistical model of the atmosphere, and models are fitted approximations that hold only until the weather does something the fit did not anticipate. An engineer sizing a link budget still has to make assumptions about the ceiling, and until now, those assumptions have mostly been guesswork. Whether knowing the real number changes what actually gets built is a separate question, and not one a formula is equipped to answer.