Low-Power Bed / Seat Occupancy Sensor Based on EMFi - Foundations on Natural and Artificial Computation - page 507

Table 3. Final results for occupancy sensor with different weight people

Weight ADC Readings Calculated
[ kg ] Empty Occupied Threshold

50 0x60 0x57 0x5C

85 0x60 0x59 0x5C

110 0x62 0x5A 0x5D

4.2 Power Consumption

The design of the new occupancy sensor for bed and couch will only be valid if the
power consumption is low significant over the average node consumption, which
results about 3 mA. Occupancy sensor consists of two different modules, the first
one detects force changes while the second one effectuates weight readings. First
module must remain permanently active while the second one is only operated
after a force change detection.
Power consumption from the force change detection module is confined to the
consumption from the operation amplifier plus two 100 kΩ resistors connecting
V CC and Gnd to generate logic level signals by means of MOS transistors com-
mutation. Overall power consumption estimated for the force change detector is
260 μA , about 10% fo the node average consumption.
The weight meter has a higher consumption when active. However, weight
measurements will only be triggered by detecting variations, so the average con-
sumption gets down. The time spent for the micro-controller to get the 150
samples has been measured in T M =74 ms . This implies a task duration of
T Check =5 T M = 370 ms , with an average consumption of I Check =10 mA ,

which is the micro-controller consumption. To estimate the average consump-
tion of the device is necessary to estimate an approximate read rate: (1) It is
considered to be detected no changes in weight, and therefore there will be no
reading, in a vacant bed or chair. (2) In a conservative way, it is considered
that an occupied bed or chair will present a mean period between detections
of variation of 10 s . It is possible to implement a state machine to establish a
maximum rate of measurement, assuming the consequent delay in the generation
of events. (3) A person may occupy a bed or chair for an average of 8 hours per
day. With these three assumptions, the duty cycle for the occupancy sensor is
DC =

·

8 h · 370 ms

24 h · 10 s

=0 . 012% and the average consumption generated by it would

be I = DC × I Check = 120 muA , around 5% of the average consumption of the

node.

5 Discussion and Conclusions

The goal of this work was to obtain an occupancy sensor for bed or couch
that could improve the contact mats previously employed in the sensor network
for a set of Ambient Assisted Living (AAL). This interest is due to the lack
of adaptability that mats presented to various factors such as hardness and
type of mattress and bed base, in the case of the bed, and the hardness of the