Door shock/opening detection - need advices

[SadE54]

I'm developing an RFM based mote that could be use as an universal bridge for alarm detectors, and can be used itself as a perimetric detector.
For shock detection I did some tests with an 3 axis accelerometer and it's ok . The ADXL362 used is really low power (270nA with Wake on motion mode !)
But for very low speed door opening , you can miss some events :-/
So I would add a 3axis magnetometer to get changes in orientation , even at low speed.
The trouble here is current consumption. For example ST LIS3DML is rated at 40uA at 20hz data rate and 10hz for 2.5 ODR . It can generate generate interrupts on threshold but there's  only basic features here. So maybe i need to wakeup my mcu to get a reading and do some processing before sleeping again.
So I could use ADXL362 + LIS3DML .
And there's the all in one solution : FXOS8700CQ from NXP
A full featured 6 axis sensor with a set of DSP features for all sort of detections. The problem is  the current draw rated at 80uA at 20 ODR . And it's ODR / 2 for accelerometers and ODR / 2 for the magnetometer.
So I'm really annoyed to make the good choices here.
I'm waiting for your advices 

[Felix]

For a door/window security device I would probably look into something like:

- shock detection (glass breaking) - this would need to withstand a strong storm/thunder vibrations
- orientation to detect when it's open - once installed this could be "reset" so the system knows the closed position. For this to really work you'd have to be able to have very high resolution, to tell when the door is just crack opened vs half/fully open. Maybe this could work in some applications, but if such a sensor does not exist then the reed sensor is the next best option and 0 power too.
- low power everything (<10uA, <1uA)

[Stark]

This is probably a bit off topic (and not as sophisticated as your requirements) but I am working on a DoorMote. Basically a reed switch which gives open and close status of my house door.

I have wired a reed switch within a MotionMote utilising the spare OLED inputs. This gives the security of a reed switch to monitor door opening as well as the PIR motion detection as backup. I am using pin change interrupts for the spare OLED input (in this case A5), which then I can monitor as a high or low to detect opening. The Motion PIR works obviously as normal and all fits in the stylish MotionMote case so meets the wife's approval.

Incidentally I am having some trouble at the moment with the coding. The pin change interupts work for a bit then after some period of idle (hours) they stop working. May be a wiring issue but wont highjack this thread any longer. Hope this provides some food for thought.

[Felix]

Not really offtopic, a motion+open event Mote is a pretty nice idea.
Unfortunately I can't think of why PIN change interrupts should stop working. Is the stopping time consistent, or random?