Finding out if someone is awake/asleep using body temperature

The temperature of the body changes during the day. So, the temperature can vary 0.5 -1 C degree. At night-sleep the temperature falls (because the metabolism decreases).

When we talk about temperature, we usually mean the temperature of the core of the body (the main organs). The temperature of the outer body (surface and limbs) is usually lower depending on the core temperature AND the environment temperature. I'm not sure if the temperature changes specifically to some parts of the body, but I don't think so.

The changes can be different at different people. However, in most the cases, the temperature will fall during sleep. People whose the decrease is very small(or not existent) might suffer from insomnia.

As we said before, the temperature is not the same for all parts of the body. However, the usual methods of measuring (for example oral temperature measurement) give a good estimation of the core temperature (a couple of tenths lower than the true core temperature).
Measuring temperature from the tips of the fingers is surely faulty (it will not indicate the core temperature). I also think the back of the neck isn't an ideal part.
The classic ways are oral rectal axillary. Of course there might be other ways if you need more accuracy.

I also found that, it says how does the temperature changes during the day. Remember that there are exceptions.
http://en.wikipedia.org/wiki/Human_body_temperature#Normal_human_body_temperature

oh, and I forgot:
If you have a temperature of a random person (without having measured him again) it is impossible to determine his state, because as we said the temperature of different persons varies. For example, a person's average temperature might be 36.5, while another's 37.5.

If you want to determine the state of a person using temperature, I guess you will have to gather some data about this person during some days. By doing that, you will know his temperature while asleep,  when he wakes up, during evening etc. That way, when you measure his temperature again, you might have a better understanding of his state (for example, he is sleeping, about to sleep or he just woke up.)
However, even for the same person, the changes of temperature varies, and depends on other things, like the mood, the hunger, illness and more. So, the temperature might only indicate a general idea of the state of the person, and with not much accuracy.

I basically need a non-invasive way like a sensor placed in the person's ear which can be used to detect when the person is awake or asleep. I've considered temperature so far but the part about having it modified for each individual person is a bit troublesome for what we're planning on doing. How much is the temperature at a place like inside the ear related to core temperature for example?

Do you think this can be made intelligent enough to be able to be used on anyone and recognise when they're asleep? If not what other physiological changes can be externally detected to know if someone's asleep?

With a short search I found that ear thermometers aren't accurate, and the shown temperature might be +-1 degree different from the true temperature. You will likely need more accuracy there, so an ear thermometer isn't ideal.

The next thing that comes to my mind is movement. Movements are reduced during sleep (but not absent).
You could put one or more movement detectors. Good places are the limbs, and (if possible) the eyelid.
The thing is, if the person stands still without sleeping, then the results may be faulty. In addition, some people move a lot during sleep, but I think most of the times you will be able to distinguish the sleeping from the awake phase. Especially if you use an eyelid sensor.

Another thing is heart rate. During sleep heart rate falls. The decrease is usually around 10-20 (for example from 70 while resting, to 55 while sleeping) but it might me more or less.
Again there are exceptions, for example if the person drinks coffee before he sleeps, the measurement might be faulty.
Well, another thing you could do is make all the measurements at once, to decrease the chance of error .

If it's possible, it would be very helpful if, before the tests, you measured the person's temperature-movement-heart rate while knowing his state. Then, at future measurements, use the previous values to determine if the person is sleeping or awake.