e-Parking Meter Management System

 

 

Home

 

Related
Work

 

System
Specifications

 

Hardware

 

Initialization

 

SAFE

 

Head
Meter
Rotation

 

Failure
Recovery

 

Energy
Consumption
Model

 

Central
Station

 

Challenges

 

Ethics

 

Conclusion

 

References

 

Documents

 

 

 

 

 

Conclusion:

 

We have developed a system to efficiently operate and monitor parking meters.  Our system assumes that there is a city-wide wireless network that can be used as a communication medium.  It also assumes that each parking meter has sensing and communications capabilities and is able to tell us (1) whether or not there is a car currently parked in that spot and (2) how much time is left on the meter. The goal was to get information about each meterís current status to some central office.  Our teamís main task was to design and implement a transmission protocol to report monitoring/sensing information from the parking meters.  Evaluation metrics include reliability, energy consumption at the meters, timeliness of transmission, and ability to recover from various failures.

 

Our team developed the SAFE (Synchronized Adaptive-Forwarding Efficient) Routing Protocol, a comprehensive routing protocol which will be used by the parking meter system in order to reliably relay information from the individual parking meters to the central station.  SAFE was designed for this particular application but can be easily adapted for any wireless network where there is a trade-off between energy efficiency and reliability.  SAFE is an on-demand routing protocol, meaning the individual meters only store routing information about the current head meter as opposed to information about getting to every other parking meter in the group.  The basic steps in SAFE are: (i) find the most efficient path to the current head meter, represented by the choice of the next meter in that path to the head meter; (ii) schedule a data packet to be transmitted before the time scheduled by the next hop determined in (i); and (iii) collect data sent by other meters, add this information to the meterís own data packet and forward this packet to the next meter at the scheduled time as determined in (ii).  The SAFE algorithm allows the user to decide on the best trade-off between reliability and energy efficiency by changing the way in which it calculates the probabilities.  The cycles are just over two minutes long, so the central station receives updated information approximately every 2 minutes.  The algorithm is able to recover from failures such as loss of a head meter, loss of a non-head meter, fragmentation, and defragmentation as previously explained.

 

Experimental Results:

 

We implemented a SAFE algorithm using Level 2 and identifying Short-Term Variations as explained above.  We ran our SAFE algorithm as follows:

 

We had meters with meter IDs 1, 2, 3, 7, 8, and 9 running.  Meters 1 and 7 were running on one board, meters 2 and 8 were running on a laptop, meters 3 and 9 were running on another board, and another laptop was acting as the central server.

 

The links between some pairs of meters (1 and 2, 3 and 8, 3 and 9, 7 and 9) were artificially made poor links by having the meters ignore 60% of the packets sent over those links.

 

Since we do not possess the sensing or coin-collecting components of the meters, the data the meters report for the presence of a car and the amount of time left is dummy data generated by our code.  However, this data changes every cycle, so those changes can be appreciated as the data is received.

 

We ran our system with this setup until the batteries were depleted.  The system ran for 181 cycles, with each cycle being just over two minutes long, making the run a total of 6 hours and 40 minutes.

 

The central serverís reliability threshold was set to 90% and the probabilities were updated every six cycles.  This means that the probabilities were increased, to a maximum value of 1.0, for a particular meter if the central server saw a data loss of over 10% for that meter during the preceding 6 cycles and they were decreased, to a minimum of 0, if the data loss was 10% or less.  For the first 114 cycles, the probabilities were increased by an absolute value of 10% and decreased by an absolute value of 10% when warranted.  Then, for the remainder of the run, the probabilities were increased by an absolute value of 30% and decreased by an absolute value of 10% when warranted.

 

CYCLES 1-114:  We had a rate of data loss of 3.7%.  Appendix C shows the number of meters with missing or outdated data each cycle due to data loss.  Appendix D shows the number of cycles for which each meterís data was received or missed.  Appendix E shows the distribution of the duration of stale information for each meter (i.e., the distribution of the length of time during which the system did not get updates for that meter).  Appendix F shows how the battery levels for each meter change over time.  Appendix G shows the range of the battery levels over time and Appendix H shows the absolute difference between the lowest and highest battery levels each cycle.  The important thing to note about the battery levels is how they decrease in tandem over time.

 

CYCLES 115-181:  We had a rate of data loss of 2.0%, more reliable than with the lower increase in probabilities during the first 114 cycles.  Appendix I shows the number of meters with missing or outdated data each cycle due to data loss.  Appendix J shows the number of cycles for which each meterís data was received or missed.  Appendix K shows the distribution of the duration of stale information for each meter (i.e., the distribution of the length of time during which the system did not get updates for that meter).  Appendix L shows how the battery levels for each meter change over time.  Appendix M shows the range of the battery levels over time and Appendix N shows the absolute difference between the lowest and highest battery levels each cycle.    The important thing to note about the battery levels is how they decrease in tandem over time.

 

Click here for Appendices C-N