Thursday, October 27, 2011

The Wireless Car Considered

In order to answer the question(s) posed in my previous post, let's go back to the automobile. Your car. Unless you are into antiques, chances are that your current car is a computer on wheels (thinking from a layman's perspective). You probably know this if you ever have taken your car in for what seems like a small repair and get hit with a gigantic bill because "the chip needed to be replaced". Or, you take your car in and they tell you they will "plug it in and run diagnostics". They download some data, study it, and presto, tweak just the right part of your vehicle. You save money.

The heart of modern automobile engines is a CPU in the engine control unit which serves as a data collection hub. Approximately 150lbs of wiring runs through your car between various sensors and the engine control unit. So when something goes wrong, it might be one of the sensors, it might be a wire, or it might be the CPU. Of course, there are many many advantages to having a CPU in your car as compared to the older technology, but this isn't the right place to go into that. What is relevant however, is the fact that all those wires are MESSY. And HEAVY. And ... complex.

Imagine the advantages that could be gained if those wires could be removed and replaced with wireless broadcasting of sensor signals to the CPU.

150lbs less weight: that's about one adult, or a few kids, or many pets, or a lot of gear. Lighter car, better gas mileage.

No wires: less mess in the cars innards (think like a mechanic to fully appreciate that one).

No wires: less complexity and cost in the manufacturing process - consider the domino effect of not having to figure out placement, routing, materials use, secondary waste generation. The car should cost less to build, thus cost less to the consumer. Economies of scale can be gained if the elimination of wires in favor of wireless becomes standard.

Store the data, compress the data, transmit the data. This is where Shannon (my final hint in the last post) comes in. Claude Shannon, whose seminal work approximately 60 years ago founded what we now know of as Information Theory. Theorems about optimal point to point communication of bits. Your cell phone calls rely on application of information theory as do your Skype calls, iPod and CD storage and other applications too numerous to count. I'll come back to that later because with our automobile scenario

There Are Challenges.

For starters, we would need to make sure that my Subaru, sitting at the stoplight next to your Lexus, doesn't tell your Lexus what to do (in theory this might be fun, but I digress). Non negotiable requirement: my car's wireless transmitters need to be picked up only by my car's CPU.

In addition, my car's CPU would have to juggle incoming signals from many sensors. There are N (meaning I have no idea how many) sensors in my car, each of them likely has a different sampling rate, transmission distances vary point to point, noise is inevitable, delay is inevitable, and there are definitely other parameters that I am not thinking of. To function properly, my wireless car's CPU must receive all the encoded data in sufficiently robust form in a timely fashion.

If you are familiar with classic information theory, you know where I am going. If you aren't, then I'll point out: Shannon's seminal work on data storage, compression and transmission didn't anticipate the kind of requirements involved in developing a functional wireless network automobile.

But there is a group of people working on this exact problem and the larger class of problems of which it is a part. And the results they come up with will effect Biology, Chemistry, Economics, Social Sciences, Engineering, and more.

To Be Continued...

 (I won't make you wait a week)

No comments:

Post a Comment