Presentation is loading. Please wait.

Presentation is loading. Please wait.

Wireless Trace Analysis. Project Goals Summary of project goals: First goal: analyze wireless access patterns Second goal: implement Markov predictor.

Similar presentations


Presentation on theme: "Wireless Trace Analysis. Project Goals Summary of project goals: First goal: analyze wireless access patterns Second goal: implement Markov predictor."— Presentation transcript:

1 Wireless Trace Analysis

2 Project Goals Summary of project goals: First goal: analyze wireless access patterns Second goal: implement Markov predictor and analyze “handoff” between Access Points (AP)

3 First Goal in Detail Analyze wireless access patterns based on several parameters such as: mobility, session and visit durations. Wireless access pattern will be extracted from wireless traces We plan to use traces at http://crawdad.cs.darmouth.edu

4 First Goal in Detail Syslog events: Authenticated, Associated, Reassociated, Disassociated, Deauthenticated We use the notion of session and visit durations A visit represents continuous roaming A session consists of a sequence of visits w/o disconnections between them

5 First Goal in Detail We show that mobility affects both session and visit durations We define stationary and mobile visits/sessions Stationary – visit only one AP Mobile – visit more than one AP (in different buildings)

6 First Goal in Detail Methodology Parser reads trace (consists of syslog messages) and maintains a state array for each client, where state can be: 1.unauthenticated, unassociated 2. authenticated, unassociated 3. authenticated, associated

7 First Goal in Detail Methodology Based on state array and access point (AP) recorded in syslog data, we determine for each user how many sessions were observed, how long they were, whether they were stationary or mobile.

8 First Goal in Detail Hypothesis As mobility increases, visit durations decrease, while session durations increase Study other metrics like hourly and weekly transfers (if time permits).

9 Second Goal in Detail Problem: track and predict the location of mobile devices as they roam or handoff from one AP to another Review existing predictors: Markov predictor, LZ predictor, PPM predictor and compare their performance Implement Markov predictor and verify the performance

10 Second Goal in detail Methodology Correlate Markov predictor accuracy with entropy (all traces) Correlate Markov predictor accuracy with trace length Time prediction using Markov predictor Duration prediction using Markov predictor

11 Second Goal in detail Methodology Improve call drop rate with training data Improve block rate degradation with training data Make a tradeoff between drop rate and block rate

12 Second Goal in detail Hypothesis Higher entropy results in lower accuracy The ratio of correct prediction increased rapidly for short traces, then increases slightly for long traces Large intervals improves time predict accuracy but degrade the precision Bigger interval led to higher duration predict accuracy

13 Second Goal in detail Hypothesis Base DR / Reserved DR ~ 5% The improvement in drop rate is correlated to the worsening in block rate in some predictors


Download ppt "Wireless Trace Analysis. Project Goals Summary of project goals: First goal: analyze wireless access patterns Second goal: implement Markov predictor."

Similar presentations


Ads by Google