Back of the Envelope

So far, so good. I've decided that the phone works pretty well as an MP3 player. I was worried at first, since it put all the music files, including all the phone's ringtones, on the same playlist. After playing with it a little, I discovered that I could make subdirectories, and it would treat each subdirectory as it's own playlist. It's not a very sophisticated solution, but it does work well for what I want to do. Namely, I like to put audio books on the MP3 player, generally putting as many CDs onto it as I can manage. This sometimes poses a problem, as when I rip the books from the CDs, it usually names the individual tracks as Track 1, Track 2, etc, so I can't put them all in the same directory. But with this set-up, I can put each CD in a separate directory, and listen to them one at a time by navigating the directory structure. As an MP3 player, it works at least as well as the dedicated MP3 player I have. In addition, the phone came with a stereo headset, where all I have to do is tap the button on the headset's microphone and it will pause the MP3 player and let me make a phone call. It also pauses the player if I receive a call, and I can hit any button on the phone to start it up again.

Which brings me to one of my favorite features, the voice activation. How I have it set up, while the headset's plugged in, all I have to do is say a "magic word" (which I choose beforehand), and it will beep and prompt me to say the name of who I want to call. I can associate a total of 25-30 phone numbers with voice commands, and for each of these I don't have to press a single button in order to call them. Unfortunately, it doesn't listen for this "magic word" while the MP3 player's going, in which case I do have to press the button on the headset's microphone.

Anyway, I finally got the e-mail working for one of the e-mail services I use. It still doesn't work with Gmail, but I figure that if I'm going to be out-of-touch with Internet for a while, I can set up Gmail to auto-forward to this other address, and access it through my cell phone. And no, I'm not going to tell you which service it is, as I'm reserving it for this particular use.

In addition, I bought a USB bluetooth adapter for my laptop. Plug it in, and I can access files on the phone from my laptop. It works well, although it's a bit slow for file transfer (~20 kbps), but that's fast enough for uploading and downloading the occasional photo, which lets me use the phone's remaining 32 MB of internal memory for pictures and reserve the Memory Stick Duo for MP3s. The picture below shows the gadgets I'm working with:

At the top is a jump drive (~$50 when I bought it six months ago), which stores 256 MB, and is useful for moving files between my laptop and a desktop when they're not sharing a network. On the left is the 32 MB Memory Stick Duo card which came with the phone (I'm using a 128 MB card now (~$50), so this is a spare), on the right is the Bluetooth adapter (~$50), and below, for size reference, is a quarter (~$0.25).

All in all I'm pretty happy. I can even upload JPEGs and MP3s to the phone to use as backgrounds and ringtones. I'm using the Imperial March from Star Wars for my ringtone right now, with a cut-out from this image for the wallpaper.

Well, I finally did it, I bought a new cell phone. The particular phone is a Sony Ericsson S710a. I've drooled over it on this blog before, and it's a really nice phone, and a decent MP3 player and digital camera. I've already upgraded the memory card (it uses the Memory Stick Duo, but alas, not the Duo Pro) from 32 MB to 128 MB. In truth, 32 MB was plenty for pictures, as it maxes out at 1280x960 pixels, but I wanted enough memory so I could take advantage of the MP3 player to carry some significant portion of an audio book with me (128 MB will hold roughly four hours, more if I were willing to sacrifice some quality). I've also gotten a look at the web browser, and being EDGE enabled, it's a real web browser, capable of displaying HTML webpages rather than just the WAP format ones. Of course, the screen's a little bit small for that, and unless you get the $25 unlimited Internet access plan, it will cost you a small fortune paying per kilobyte. What I really want to do is check my e-mail using it, but while it has POP3 capability, it doesn't play nice with Gmail. It can access the Gmail website directly, but again, the additional data overhead could get expensive. If I hadn't gotten comfortable with Gmail, I'd consider getting an account with one of the other services which the phone does work well with. I still might, just to see whether I like how it handles e-mail and whether I want it badly enough to move my MIT alum e-mail forwarding to yet another account.

Anyway, the real reason I got this phone was the camera. I was tired of coming upon something and thinking "I really ought to get a picture of this for my blog" and not having a camera with me. Now, even with this phone, the camera's not great as digital cameras go, but it's good enough for the small pixel size images you usually put on blogs.

I like using online maps when I'm trying to figure out how to get somewhere. I enter my starting point and my destination, and it outputs the optimal route to get there. Well, except for that "optimal" part. I'm not sure how it decides what route to give you, but whenever I'm trying to get somewhere around Boston, it directs me to I-95. I-95 is the last route you want to take anytime between 7 am and 6 pm, and in the Boston area, there are at least a dozen back roads you can take instead. So what I want to know is are there any online map programs that will provide alternate routes on request.

Doc is hypothesizing about a toy:

OK, let's just suppose, as a purely abstract exercise, that you knew someone who had just moved into a new office building with lots of empty space. Also suppose (this is purely fictional, you understand) that this person was toying with the idea of constructing a remote control vehicle that would work by connecting to the company's wireless network (of course this would be an unconscionable misuse of corporate resources) so that he could put a USB video camera on it and control it from his desktop to run around the office after hours, after all the suits have left. How would you recommend this hypothetical person go about accomplishing this hypothetical goal?

This actually sounds like a fun project, and it shouldn't be too hard. It's not difficult to reverse engineer an RC car's controller. In fact, I've done it before, as an engineering project in college, but as that was eight years ago, I don't remember it too well. Hmm, as I recall, it was fairly straightforward, using a PWM to control the speed, and a straight analog voltage (I built a simplistic D/A converter) to control the steering. I used an HC11 microcontroller to run it, guided by infrared sensors which read the car's position on a track it was supposed to run, but you could replace the HC11 with something with a WiFi link. I'm sure Doc can do it, even if he is a programmer rather than an engineer. Hypothetically, anyway.

Can you spot what's missing in this article?

Toshiba Research Europe has used the science of quantum cryptography to transmit voice and video over a secure fiber link that is protected by the laws of physics. The demonstration is significant because it shows that the single-photon encryption technology is not only compatible with real Internet Protocol (IP) traffic but also robust enough for deployment on commercial fiber networks.

The system was shown to financial institutions and government representatives in London last week by scientists working at Toshiba’s Cambridge Research Laboratory.

Toshiba’s “Quantum Key Server” can generate up to 100 quantum (single-photon) keys per second, enough to encrypt each video frame with a separate key. In addition, it features an automatic management system that continually monitors and adjusts the system’s optical path length to allow it to operate continuously without any need for user intervention.

If you remember the things I talked about in my previous posts on quantum cryptography and RSA encryption, you should be scratching your head wondering something right now. There's a critical piece of information missing, and it's nowhere to be found in the article that I quoted. Go ahead and follow the link and read the article. I read it three times before I was convinced that they really had left it out.

Nowhere do they state what the length of the key is! Nowhere! How in the world can anyone write an article, where they make a point about how fast this new system can send the quantum information, without telling you the data rate? Talking about sending 100 keys per second is meaningless if they don't say how long the key is!

Okay, let me calm down and explain. Recall that, using quantum key distribution, you can share random bits between two users, and these bits form your key. One user uses his key to encrypt his data, which he then sends over a public channel, and the other user receives the encrypted data and uses his key to decrypt it. Since they have the same key, transmitted securely, they can share encrypted data confident that no one else can decrypt it. The trick here is that quantum key distribution allows you to transmit the key securely, but it's slow. Very slow. For one, you end up losing three-quarters of your data up front, and the more secure you want to make the transmission, the more data you have to toss aside. So some day it may be possible for quantum key distribution to distribute the key at somewhere around one-quarter the speed of the open, unsecured data line. Someday. The technological difficulty of transmitting and detecting one photon at a time, which is necessary for quantum key distribution, makes it even slower. The highest data rates are somewhere around 10 kbps (10,000 bits per second), as I mentioned in a previous post. For complete security, your key has to be as long as your data, meaning it will take longer to send the key than to send the data.

Now, you can use a key shorter than your data length. It won't be as good of an encryption, but you can encrypt data with keys which are much smaller than the data you're encrypting, just using the same key over and over. The problem is with a short key, a hacker who intercepts the encrypted message can just keep trying possible key values over and over until he gets an output that makes sense. For example, you can encrypt a phrase with an 8-bit key, but it will take someone no more than 256 tries to find the right key, and once he finds it, he'll know, since the decrypted phrase will make sense. Now if your key is as long as your data, then no key he tries could differentiate it from any other phrase of the same length.

So let's say you have a quantum key distribution system running in parallel to the data you're sending, but much slower. The best way to use it is to produce a key, say 100 bits in length, and use it to encrypt the next 100,000 bits of data. Then, when you produce another 100 bit key, you use it to encrypt the next 100 kb block of data. This works well when you have a 10 kbps quantum key distribution system and a 10 Mbps open data channel. This is what's being done in this article. They produce a key, encrypt some amount of data, then produce a new key and encrypt then next block of data, and so on. They can do this 100 times a second, enough that if they're transmitting video, each frame can have its own encryption. (This is actually an understatement, as video framerates are typically somewhere between 10 fps and 60 fps.) There's just one tiny problem--they don't tell you what the key length is! Without that information, talking about the number of keys is meaningless. The key could be only 10 bits long, in which case this is an exceedingly slow quantum key distribution system with a 1 kbps key data rate. It could be 100 bits, which means it's about average as quantum key distribution systems go at 10 kbps. Without telling you the length of the key, they aren't telling you the key distribution data rate, which is the number of keys times the length of the key. What's the point in talking about the number of keys if they don't tell you the key length?

I had to do a bit of hunting around the internet until I could find an article which said what the data rate is. A surprising number of articles neglected to say. This one finally shed some light on it (and it still didn't give the information the proper emphasis):

Their system is capable of generating 100 quantum 'keys' every second. This is fast enough for every individual frame of video to be protected by its own encryption. "This makes the system highly secure," says Andrew Shields, who leads the Cambridge team. "It would take an enormous computational resource to crack this frame by frame."
...
The Toshiba system creates keys made of 256 'bits', where each bit is a photon speeding along a fibre-optic cable. A photon represents either one or zero depending on whether it arrives slightly early or late at its destination. By passing a series of messages between the sender and receiver, both can arrive at a secure, mutually agreed key.

So, it's a 256 bit key, transmitted 100 times per second, a distribution rate of approximately 25 kbps, roughly 2.5 times the previous state of the art. Not bad, but do you think they could bother making sure that information is emphasized next time? I know I should blame the reporters for missing that crucial piece of information, and I do, but I think that if Toshiba had emphasized it, they might have picked up on that fact.

Is time travel possible? Amal Dorai of MIT intends to find out:

Attention, time travelers: Amal Dorai hopes you enjoyed the party he's throwing this weekend. Dorai, a student at the Massachusetts Institute of Technology, is hosting a Time Traveler Convention on campus this Saturday. Make plans now, because it's the last such party.

"You only need one," he said. "The chance that anybody shows up is small, but if it happens it will be one of the biggest events in human history."
...
Dorai only asks his guests to show proof they come from the future: Bringing the cure for cancer, a solution for global poverty or a cold fusion reactor would suffice.

In case MIT is long gone by the time a time machine is invented, Dorai's invitation includes geographic coordinates for the East Campus Courtyard (42:21:36.025 degrees north, 71:05:16.332 degrees west).
...
The convention starts at 8 p.m. For dramatic effect, time travelers are encouraged to show up at 10 p.m. sharp. In between, revelers will take in a lecture on time travel by an MIT physics professor and listen to student bands belting out time-themed songs.

So don't forget, time travelers should show up at 10 pm Eastern Daylight Saving's Time on May 7th, 2005 AD, at 42:21:36.025 degrees north, 71:05:16.332 degrees west. If I could time travel, I'd go. I might go anyway--it's at MIT tomorrow evening, which shouldn't be too hard for me to make. Time travel's probably impossible, but if someone does come, I'd hate to miss it.

This sounds like a cool idea, from Doc Rampage:

I have this fantasy called "programmer appreciation day". This is the day, once per year, where we line up all the computer programmers that have made our lives miserable through their gross stupidity. Everyone gets to go down the line slapping whichever ones they think deserve it.

Have I mentioned that Doc's pretty curmudgeony?