Development of a GIS based software for real time noise maps update andrea cerniglia
Scale and sum Main idea of the system is to scale a set of pre-computed noise maps according acquired noise data, and sum them together in order to obtain the map of the whole area.
Dynamap
Scale and sum
Communication between monitoring stations and data collecting unit is not limited to the transmission of the acquired data but it is, in fact, a two-way communication
Thanks to this bi-directional communication is possible to set-up and update the monitoring station remotely
Delay between Timestamp arrival and time The calculation of this parameter permits to detect the amount of queued data (i.e. data transmission channel fault)
Delay between Data Sent and ACK received The calculation of this parameter permits to detect problems related to bandwidth of transmission channel (i.e. insufficient bandwidth)
The scaling and sum of 10 maps with 300,000 points each involves: 3,000,000 sums (10 x 300,000) 3,000,000 divisions 3,000,000 antilog 300,000 sums with 10 terms per sum 300,000 log calculation 300,000 products
In order to optimize the ‘calculation power compsumption’ needed for this enourmous amount of operations, some tricks are possible to simplify complex operation like log and antilog, to faster operation.
This can be done by pre-compute una-tantum all the possible antilog in the acoustic range (20 to 120 dB) and by storing them in a table so, pratically, the 3,000,000 antilog calculation consists then just in a fetching peration on a table, like in an abakus.
Also the division by ten and multiplication by ten can be semplified to a simple sum by using logaritms properties
After the implementation of these ‘tricks’, the scaling and sum of 10 maps with 300,000 points each involves ‘only…’ 3,000,000 sums (10 x 300,000) 300,000 sums with 10 terms per sum 300,000 log calculation
Similar approches are used for the log calculation process and for the sum process. More in details, the ‘trick’ is to pre-compute all the numbers in the intervals which gives each 10 log (x) possible value, and just fetch the correct value in a pre- computed table.
System configuration
With the above mentioned configuration and with the implementation of the algorithm for computation time optimization, the time needed for scaling and sum 10 maps of 300.000 points each, on a medium level machine, is less than 5 seconds. The most critical aspect in terms of time consuming is related with the disc access time (for retrieve data and store results) which can improved by means of SSD insead of standard mechanical disk (in progress).
Databases · Dynamap_Mon_Data · Dynamap_Mon_Results · Dynamap_Basic_Maps · Dynamap_Maps_Results · Dynamap_Access_Privileges · Dynamap_Administration
Dynamap_Mon_Data database: location table
Dynamap_Mon_Data database: Values table Noise values are stored in Bel instead of decibel, once more to deal only with integers (instead of real numbers) and thus improove memory allocation and operation speed
Dynamap_Mon_Results database The system automatically compute hourly Laeq values for each monitoring station and store them in a database
Dynamap_Basic_Maps database: Values table
Dynamap_Maps_result database: Values table
At the end of the ‘scale and sum’ process, all the computed values are avaible to another part of the system (still under development, ready to be released in autumn 2016), able to produce web-based noise maps on demand, and able to to show them to citizens.
Thanks for your attention andrea.cerniglia@accon.it