That depends on the system you want to use.. you have a few choices..
You can identify each node and each device uniquely..
device 10 is node 1, node 2 is "device 20" node 3 is device 30...
temperature is device 01, pressure is device 02, humidity is device 03..
Temp sensor on node 1 is "device 11", Temp sendor on device 2 is 21, humidity on node 2 is 23.. pressure on node 4 is 42.
... this scheme also leaves room for expansion.
You could choose to use data authentication..
The acceptable ranges for your temp , pressure, and humidity are going to be pretty different.
--Of course this depends on your application.. something in deep space, or near a volcano may have very different needs than a backyard monitoring station.
the type of data is also very different.
Temperature is only going to be in "degrees". If you go off the normal Fahrenheit / Celcius scales.. and use something like Kelvin.. then your valid data is around 270 degrees - 375 degrees
of you can use the numbers we are used to.. but consider unique cases when temperature is equal to the humidity measurement.
Humidity is going to be a percentage, valid between 0-100 / or 0.00 - 1.00
Pressure is going to have a narrow range in bars : 870 mb - 1080 mb ( probably narrower.. these are record numbers)
You could also use a polling scheme, similar to token-ring.. where each unique device is polled at a particular time interval. Only when a sensor is queried - will the data returned be valid. Then your controller 'knows' which device is being queried, and therefore, what the data means.
There are probably other ways.. this is just what I came up with quickly for you.
misc. knowledge.. trivia.. google