Unless you have specific requirements for the master, there is no need for having a master to collect the data from the slaves and send it to cloud. You can have the same code running on all the Spark modules. Each spark core can get the data from the PIC and send it to the cloud using Spark.publish. Please note that you need a Web page to pull the data and take some possible actions. You can have a look at the tutorial on Spark.publish at the link
Hope this gives you a good starting point