Donate?
I file this under, "Ideas I should have patented myself and profited from." Long live capitalism!

Smart Cards
I have a friend who's already programming smart cards to replace casino chips entirely. It's already been pilot tested, and from most indications customers have been quite happy with it. There's no RFID chip though, you actually place the card into a reader at the table. The thing that boggles me, and that I have a hard time believing is a good idea, is that the value of the card is stored completely on the card itself (encrypted and all, of course). There is no back end "accounting" app that knows how much is on each card.....

Donate?
I'm all for profit, jimothy. ;)

[Begin Patent Rant]I just think it works better with patents reserved for the use they were originally intended, to promote innovation. Software and business practice patents only stifle innovation as far as I can tell. How often do you think software patents are actually used to protect and inventor's investment in research and development by giving that inventor time to market their invention? How often are they used by non-inventors to raise money purely through litigation? [End Patent Rant]

With that said, I'd have probably given away a bare-bones version of the software too and then made my money in niche consulting configuring and customizing the systems. That way I stick to what I do best and don't have to become the manager of a company tied to a single product.

Smart Cards
This is a recurring question I have for electronic wallets in general. What is the benefit of storing currency on the card as opposed to using the card as just an ID and maintaining the balance in a central server?

I assume the benefit is the savings for the Casino in not having to cable the readers to the network (or support and secure wireless links). But then how are they doing accounting at the end of a shift? Do they go around with a dongle or token card and transfer the tokens from each table or slot machine?

The most obvious cost is the risk of loss of currency by the casino customer. For instance, what happens if I damage or loose my card? If the Casino retains a way to invalidate an electronic token, they could cancel and reissue the card for any amount not found spent at a table or slot machine, but then the system would again have to be real-time and that means connecting the slots and tables into the network.

There's a hidden cost as well. In a centralized system, we only ever transfer authentication information and an authorization for a transfer. We have the opportunity to monitor this system for intrusion, advantage Defense (Alice and Bob). Someone changing cash value on a card could leave the building, work on the card at their leisure and return, advantage Offense (Mallory in this case).

More on the Adventures of Alice and Bob

RFIDs and Interference
I don't know about the tiny RFIDs they have nowadays, but at least with the card RFIDs I use for building access, if you have two that are too close to each other, you won't get a good read from it.

If the RFIDs are all transmitting at the same frequency (which I assume they will be), won't they all be interfering with each other?

RFIDs and Interference
Your intuition is excellent. The protocols used to keep tags from talking over each other are called anti-collision protocols. Many older RFID tags were created when the cost of anti-collision logic was too great for applications where the chance of collision is low. In the case of building access cards, the read is very short-range and two cards aren't often in the field.

The other problem when many tags are expected to be in the field at once is how to find one tag out of many and tell the others to shut-up. This is called "singulation." Think of it like a turnstile which singulates people through a door.

There are three main variations on anti-collision and singulation protocols in use at the moment. Aloha, Slotted Aloha and Binary Tree protocols.

"Aloha" developed in the 70s by Norman Abramson of Aloha Networks in Hawaii for packet radio communication. Aloha originally inspired the Ethernet protocol and in its Slotted variation is still used for satellite communication as well as for ISO 18000-6 Type B, and EPC Gen 2 RFID tags.

Aloha itself is as simple as an anti-collision procedure may reasonably be and doesn't really include the concept of singulation in the usual sense. With this procedure, tags begin broadcasting their IDs as soon as they are energized by the field of the reader. Each tag sends its entire ID and then waits for a random or pseudo random period of time before broadcasting again. The reader simply receives the IDs, depending on chance to ensure that each tag will eventually happen to broadcast during the quiet period of all other tags. The reader doesn't respond to the tags in any way. The advantage to this procedure is speed and simplicity. Tag logic is minimal and with such a low protocol overhead, this is the top performer for read rate where only a few tags are present.

Slotted Aloha takes that protocol and adds "slots" which are periods of time during which tags may speak. Instead of a random interval, tags pick a random slot to communicate in. Once the reader catches a tag transmitting without collision it singulates that tag by telling it to use a special slot to communicate. Other tags won't use that slot unless told to do so.

Binary Tree protocols step through the possible bits in a tags identifier, one bit at a time, essentially saying, "Anyone who has '1' as their next bit stay with me, everyone else go silent." repeating this until only one tag is left.

For more info, look at the EPC UHF tag specs and here's a nice link on Aloha and Slotted Aloha.
Also, if your really patient, our book should be coming out this spring, and there's a chapter which includes singulation and anti-collision. ;)