You are not logged in.
If you Google, "the universe is unkind to electronics", you'll quickly understand why it is that electronics are not long-term reliable control systems for sophisticated systems such as navigation, communications, life support control and monitoring systems, and general purpose computing. The radiation environment of space is constantly fusing together or punching holes through the logic gates of the highly integrated circuits. However, the statement, "the universe is unkind to electronics", or "the universe is hostile to computers", is only partially true. The universe does not affect photonics-based compute devices to nearly the same degree, nor in the same ways, because they operate on very different principles. Electronics-based computing uses highly regulated low-voltage direct current to make electronic compute devices as efficient and compact as they are. Photonics-based computing uses coherent light sources, the crystalline / physical structure of the circuitry, and dopants to diffract light in useful ways, in order to do what digital electronic computers do.
A photonics-based CPU would consume about 1/1,000th as much power as current Silicon-based semiconductor CPUs. If the entirety of the compute device operated upon photonics-based principles for data processing / transfer / storage, as opposed to the current rather clunky electronic-to-optical-to-electronic interfaces, then it would consume about 1/1,000,000th as much power. This would enable an AI with the same sophistication as a human brain to operate on the same total input power / wattage as a human brain. Replicating the functionality of a single human brain requires over 20MW of electrical power input to power Silicon-based semiconductor CPUs, as well as the supporting electronic circuitry for data transfer, storage, and I/O. The physical size of these Silicon-based devices remains far too large to fit inside something approximately the same size of a human's head. This limits its general utility and applicability to AI-enabled robots, because a computer that consumes 20MWe is a stationary device similar in size to a small building.
AI may still be very useful for a small number of research scientists to evaluate in a research lab with ample electrical power and cooling capacity, but what it can feasibly do, and for how many people, is drastically limited by its incredible power consumption. Until more people can take advantage of what a Artificial General Intelligence (AGI) has to offer, AGI is a curiosity for scientists to poke at. To advance AI-enabled computing beyond the current paradigm of ever-increasing energy consumption for increasing compute throughput, eclipsing global civil aviation in terms of energy consumption, we must make radical improvements to CPUs, data transfers, and memory storage that only optical / photonics-based computing can deliver.
Electronics were a good start, as it pertains to integrating the functionality of various discrete components. We ought to know that they are merely a stepping stone to something far more reliable and less energy-intensive to both make and use. That is the tantalizing promise of photonics-based computing.
The Universe is Hostile to Computers - Veritasium
Apart from photonics-based CPUs, optical data storage is another very important part of photonics-based computing.
Scientists Record the Human Genome to a 5D Memory Crystal - by Dr Ben Miles - Physicist
To test the crystal’s potential to be used as a repository for important genetic information, the researchers inscribed the full human genome into it. Each of the approximately three billion letters that comprise the genome was sequenced 150 times to ensure it was in the correct position.
The crystal was also inscribed with a key to ensure that whatever intelligence discovers it in the future – human, machine, or alien – will know how to use the information it contains. The key shows the universal elements (hydrogen, oxygen, carbon and nitrogen), the four bases of the DNA molecule (adenine, cytosine, guanine and thymine) with their molecular structure, where they’re placed in the DNA double helix, and how genes are arranged on a chromosome.
The genome-containing crystal is currently being kept safe. It’s in a time capsule in an ancient Austrian salt cave, overseen by the Memory of Mankind archive.
* Memory crystal was developed by the University of Southampton’s Optoelectronics Research Center (ORC) using nanostructured glass
* Memory crystal is made from quartz glass, thermally stable from freezing temperatures to 1,000C, and roughly the same size as a quarter
* Stores 360TB of data, using a femtosecond laser capable of etching / encoding 20nm-sized features into the crystal
* Can theoretically retain the stored data for up to 300 quintillion years
* Highly resistant to galactic cosmic rays and other forms of radiation
* Bytes of data are stored using height, length, width, orientation, and position (5 dimensions of freedom to express digital data; 2 optical dimensions and 3 spatial coordinate dimensions)
The human brain can store approximately 2.5 petabytes of information.
1 petabyte = 1,000 terabytes
That means a machine wishing to store the same total quantity of data as a human brain would require about 7 of these memory crystals.
Star Trek's Commander Data could store 800 quadrillion bits of data, so his positronic matrix would require 278 of these coin-sized quartz crystals.
The remaining question is how to best power them, and how to maximize data transfer rates. Microsoft's Analog Iterative Machine (AIM) computing platform has truly astonishing data transfer / throughput rates, because it can use light with every color in the rainbow to process data.
Beyond that, we then need the ability to communicate or share processed information over networks using optical methods, so that we're not constrained by the bandwidth limitations of radio waves. This typically involves lasers. That said, radios are not going away any time soon. There are certain communications or sensory tasks for which radios, radars, and sonars can do that photonics-based communications devices, such as lasers or lights, simply cannot. There is no such thing as a good laser-based communications solution for transmission of data through clouds, the ground, or water, for example.
However, a photonics-based solution for processing of radar return data would drastically reduce the size and total power consumption of a radar sensory device, as well as its cooling requirements, because it could enable algorithms which clean up the data so that lower transmitted power is required to build a high quality radar image. Highly sophisticated mathematics is used to distinguish between false returns or noise or terrain features and actual moving targets such as birds or other aircraft. The more sophisticated the processor, the more the radar can show you without increasing its power output level.
The SAR spy satellites operating in medium orbits above the Earth are not so powerful because they pump greater wattage through their antenna. They apply exceptionally sophisticated post-processing algorithms to returns data, sometimes using AI-enabled server farms back on Earth, and have a highly sensitive antenna, which is how and why these rather small X-band radar satellites, using maybe 1/10th of the emitted power from a fighter jet's X-band radar, can very clearly "see" stealthy ships or aircraft in flight from hundreds of kilometers away. If they had onboard photonics-based computing, they would not require any server farm back on Earth, and would simply send finished data products back to Earth.
While there are many obvious military applications for photonic computing tech, there are also civil Earth observation and geolocating uses, such as being able to pinpoint exactly where on a planetary surface some wayward person, aircraft, or vehicle happens to be, and to provide that data in a matter of seconds. The tech can perform long term monitoring of weather and climate, because you cannot mess up the sensors or computer to nearly the same degree as electronics-based equivalents. At the present time, we cannot accurately model atmospheric water vapor distribution or temperatures, due to our inability to distinguish signal from noise with the required degree of accuracy. All the manipulation of the data is required because there's no sense to be made of it, but the end result has been no agreement on what data to use, and none of our climate models can back-cast to accurately model historical temperature records. This is highly problematic for both weather and climate. If you cannot accurately model the past, then you cannot accurately predict the future, either.
A vastly more powerful photonics-based computer aboard a satellite could show us water vapor volumes and density and distributions in near-real-time, decisively settling the science behind climate change in a matter of a years, because increasing warming would show an unmistakable relentless increase in the total volume of retained atmospheric water vapor- a directly measurable physical quantity (when the required computing power and precision is available) that is only explainable by a general increase in retained global thermal energy or photo-based evaporation, which is again only explainable by an increase in the total amount of trapped or absorbed IR wavelength photons (input energy from the Sun or other terrestrial sources). Trace atmospheric gases such as CO2 and Methane produce highly variable and localized atmospheric behavior, but water ALWAYS evaporates at a faster rate as thermal energy increases. There are no two ways about this. If there are 10 gigatons of water vapor in the atmosphere during Year 1 of measurements, 11 gigatons during Year 2, and 12 gigatons during Year 3, after perhaps 10 years we can call that whatever we wish, but it's a definitive pattern and sign of things to come.
At a more mundane level, being able to instantly find someone who has gone missing is an invaluable tool for preserving human life and curtailing human trafficking. If you're lost, look up at the sky, we drop a pin on your location, and then we send someone to come get you. Governments will always be tempted to abuse this capability for nefarious purposes, so that is why anyone the government is looking for should be public knowledge, and subject to civil oversight.
For science missions to other planets, dramatically more detailed maps of the planets would be very nice to have, but the sheer volume of data generated and the processing power required are well beyond the state-of-the-art.
Offline
Like button can go here
This post is reserved for an index to posts that may be contributed by NewMars members over time.
This new topic has significant potential for development. On the other hand, the subject matter is somewhat advanced.
NewMars members can contribute by watching for announcements/articles that relate directly to the topic and posting links here.
(th)
Offline
Like button can go here
To be clear, NASA's CALIPSO (lidar-based; end of mission was August 1, 2023) and CloudSat (radar-based) satellites have already performed some of this atmospheric data gathering mission addressed in my first post, but the quantity of data gathered relative to the quantity of data required to produce near-real-time spatial and volumetric measurements falls within two entirely different classes of data sets. CALIPSO and CloudSat are looking at the atmosphere through pinholes, relative to what I'm proposing, which is to effectively monitor the entire atmosphere of Earth at all times, in order to build a literal near-complete picture of what is going on. The old saying that a picture is worth a thousand words is absolutely true.
I want to use photonics to take all these good science experiments, which do provide useful bits of data, and convert them into mature continuous 24/7/365 monitoring systems so that we're no longer so much as making educated guesses, because we have continuous complete coverage monitoring, thus no ambiguity left regarding what is happening, where, and when. We no longer need to extrapolate or homogenize data sets when we have continuous recordings. People will still argue over interpretation of the data, but there should be no argument left about what the actual measurements are.
For aircraft, a completely photonics-based computing solution would mean we have all-glass cockpits and sensors / inputs that are entirely independent of onboard electrical systems. They will use so little electrical power that something the size of a D-cell battery is entirely sufficient to power the onboard compute / sensor / communications suite. The possibility of an alternator or circuit breaker failure, and accompanying power surge, disabling all or most of the flight instruments is then effectively nill. The electrical system is only required for lights, autopilot and electric trim servo systems for aircraft so-equipped, and possibly the flaps (I would argue that flaps for small aircraft should be entirely manual). Propellers and landing gear are either fixed (fixed pitch props, spring-based landing gear) or hydraulically actuated (brakes, oleos, constant speed propellers). The trend towards making every system aboard the aircraft electric or electronic, however inappropriate, is finally arrested. Overall reliability is improved by using more reliable traditional control methods or reducing the effects of a single point of failure, in the form of the aircraft's onboard electrical system.
The numerous and repeated failures of electronic parking brake devices immediately comes to mind. There was no net benefit to aviation or motor vehicle safety from implementing such devices, despite objectively false claims to the contrary. It was solely intended to keep electrical engineers employed by doing things with electric / electronic devices that they knew or should have known would be detrimental to reliability and safety. The reason almost all parking brakes were manual devices is not that it was impossible to include an electric servo to actuate the parking brake. Doing that was adding a lot more complexity, potential failure points, and cost. The objective idea behind incorporating a parking brake into an aircraft of motor vehicle is not to drive up the cost of the vehicle it's attached to, it's to make sure the vehicle doesn't roll when the parking brake is engaged. We devised a simple and reliable way to do that, proven to work over many decades of operational use. Unscrupulous engineers came along and found a way to milk a job out of complexifying a silly parking brake by adding electronics. Every automotive manufacturer and many light aircraft manufacturers are now employing an extra electrical engineer to design and maintain silly parking brakes- a purely mechanical engineering problem that was solved a century ago. It's pointless whiz-bang gadgetry to entertain the brains of electrical engineers by extracting money from customers who had no choice in the matter, and would always choose a simpler / more reliable / cheaper mechanical parking brake solution if they understood the cost-benefit analysis of both systems.
As for navigation in adverse weather, exceptionally detailed topographical maps of the entire world can easily be stored in the vehicle computer using crystal-based storage.
Piston engines equipped with laser-based ignition systems and photonic computers would ensure that the fuel ignition system is about as reliable as practical, as well as improving fuel economy through better combustion efficiency. There's no practical way to light off the fuel-air mixture at multiple points inside a combustion chamber using electric spark plugs, but lasers make multi-point ignition a realistic option for hybrid spark and compression ignition engines using spark-based ignition optimization to reliably start, regardless of temperature, and to ignite the intake charge even when it's too lean to ignite using traditional spark plugs. The computer itself could be mounted directly to the engine, possibly even inside its exhaust manifold, to act as a self-powering control system using the heat / light from combustion to deliver input power. Eliminating alternators and batteries would represent a step-change for total vehicle efficiency, reliability, and weight reduction.
For an aircraft battery and alternator combination, you're typically looking at 50lbs to 75lbs of dead weight that must be carried aloft. That is why early aircraft didn't have batteries or electric starters- the power-to-weight ratio of early piston aircraft engines was so poor that there simply wasn't sufficient weight and volume allocation for such systems, so the engine was manually started on the ground to eliminate the associated performance penalty. For a land-based passenger vehicle, the weight penalty figure is closer to 100lbs.
Adding a bunch of heavy electric / electronic gadgetry is a step in the wrong direction. We should be looking at ways to eliminate as much weight and complexity as we can reasonably get away with, because all of it ties back to overall transport efficiency.
Believe it or not, there are even laser-based audio systems that generate sound by using laser frequencies that strongly absorb into atmospheric water vapor to produce sound / vibration, thus audio systems do not need to be electrical / electronic, either.
The point behind all of this new photonics-based tech is radical efficiency and reliability improvement over more primitive electrical / electronic systems, to the point that many traditional vehicle components which generate or transmit power are no longer required, because a comparatively trivial amount of power in the form of photons is being consumed to perform the functions of electrical and electronic systems. A modern motor vehicle consumes hundreds of Watts of electrical power for all the various sensors and systems onboard. This power comes from an alternator that must be spun by the engine, and is typically not very efficient relative to an electric drive motor. If we could eliminate the alternator and the battery and the tens of kilograms of Copper wiring, replacing them with a system that might weigh a few kilograms in total, then both the materials demand and wasted energy input is considerably reduced. It's not written in stone anywhere that we need some of this stuff, but headlights are pretty useful, radios and navigation systems tell us exactly where to go and what to avoid, digital engine control is the only reason an engine with the size and weight of a lawn mower engine can produce 100hp+, and there is no real reason to think that manufacturers will quit adding "more stuff" to their designs, so a serious overhaul of the fundamental power and control systems is the only way we're going to provide that without already unsustainable cost increases. The electronics in a motor vehicle now represent 50% of the total cost of the vehicle. Something must be done to reduce the ever-growing logistical tail associated with all these fantastic machines we've built.
Offline
Like button can go here