

eParking Meter Management System



Energy Consumption Model:
We are basing our energy consumption model on
the one proposed by Laura Marie Feeney, a linear model governed by the
following equations for sending and receiving broadcast and pointtopoint
messages:
The wireless cards we are using for our prototype do not have the capability to adjust the power level they use when sending. However, other cards do. In a system using such cards, the power level for broadcast send could be set lower than for pointtopoint send, since broadcasted messages only need to be heard by the meter’s neighbors and pointtopoint messages need to be heard by the access point, which is usually farther away than a neighboring meter. Therefore, we are pretending to use two power levels, one enough for a message to reach two meters down the line and the other one enough for the message to reach the access point. There can be further energy gain by assuming a continuum of transmission power levels as demonstrated in the PARO (PowerAware Routing Optimization) designed by Gomez, et al. However the implementation of such complex protocols is beyond the scope of this project, especially since it cannot be realized with our resources.
In our implementation, routing is done at the Application Layer which means all broadcast packets are processed regardless of the destination. For this reason, we chose to make the transmission from Head Meter to Central Station a point to point transmission. This way the surrounding meters will not have to process these larger packets and thus conserve some energy. Moreover, sending pointtopoint takes advantage of retransmissions at the MAC Layer, reducing the probability that the packet from the head meter will get lost. The disadvantage is that the Head Meter will have to spend more energy since the cost associated with pointtopoint transmissions is higher than that for broadcasts.
We are estimating that the access point can be 5 times as far from a meter than the meter two spots down the line. Therefore, since the power required to transmit to a distance x is proportional to the area of the sphere of radius x, transmitting to 5 times the distance requires an increase in power to the order of 25. As a result, the power required for a pointtopoint send is 25 times the power required for a broadcast send, and thus the energy consumed by a pointtopoint send is 25 times the energy consumed by a broadcast send. Therefore, we are using Feeney’s equation for broadcast send and broadcast receive, but we are multiplying her equation for pointtopoint send by 25. Our meters will never receive a pointtopoint message, only the central server does.
We accelerated the rate of depletion by multiplying each equation by 10,000 so that battery depletion could be appreciated in a relatively short amount of time. We started with the batteries at 11050J, which is the energy contained in a standard Lithium Ion AA battery.
A run of the SAFE algorithm indicated that the algorithm consumed approximately 1657.5 Joules per Hour, which in a real system translates to 0.16575 Joules per Hour.
This estimate for the rate of energy consumption is lower than what it would be in a real system. This only includes the cost of operating the wireless card for sending and receiving. As mentioned before, it does not include the cost of keeping the card idle. It also does not include the cost of operating the meter, including the sensor. Moreover, this number was calculated for a group of 6 meters; the higher the number of meters in the group, the higher the overall cost of sending and receiving since there will be more update packets and the data packets will be longer. 