amankaraj / indoorlocalization Goto Github PK

Given:
A set of data rows.

Condition:
Person is moving, so we'll be getting the signal strengths for each point, each second.

Idea:
We will be able to identify few of the points using the data, but not all of them, but we should be able to use the history, maybe last 60 seconds to make a wild guess.

Approach:

??

wirelessdicussiion.docx

Given data in below form:
[32197,7838,-1]
Loc: 40.90986093 -73.11787836 30_19_42_57
GSM RSSI: -61
EDGE Cell 0CID: 6232 LAC: 32197 rssi: -83
EDGE Cell 1CID: 6233 LAC: 32197 rssi: -77
EDGE Cell 2CID: 5338 LAC: 32197 rssi: -97
EDGE Cell 3CID: 6231 LAC: 32197 rssi: -101
EDGE Cell 4CID: 7839 LAC: 32197 rssi: -99

Condition:
Loc represents the Latitude Longitude locations, followed by date_time
Other lines represents Cellular tower id and how much signal strength we are getting from each tower.

Issue:
The count of towers from where we are getting the strength varies, at times there are 3 towers at times there are 4.

Given:
For a single location we have the data from cell towers in the below format:
EDGE Cell 0CID: 6232 LAC: 32197 rssi: -83
EDGE Cell 1CID: 6233 LAC: 32197 rssi: -77
EDGE Cell 2CID: 5338 LAC: 32197 rssi: -97
EDGE Cell 3CID: 6231 LAC: 32197 rssi: -101
EDGE Cell 4CID: 7839 LAC: 32197 rssi: -99

Condition:
Consider the case when we move to a different location.

Issue:
We might end up getting same signal strengths from different set of towers. Key point is, at any given point set signal strengths from towers would be mapping to a particular location. So we will have to find a way to include pairs in our training data set.

Approach:
??

Implement a function to apply Naive Bayes on a List of List of values, where each value would be a float.

Parser not working for big data file because the number of signals are varying. They range from 0-4 and 0-3.