VTC2005 fall1 Understanding the effect of environmental factors on link quality for on-board communications Irene Chan, Albert Chung, Mahbub Hassan University of New South Wales Kun-chan Lan, Lavy Libman National ICT Australia (NICTA)
VTC2005 fall2 About NICTA A national research institute funded by Australia Government Our research staff includes –regular researchers –contributed staff from major universities such as Australian National Univ., Univ. of Sydney, Univ. of Melbourn, New South Wales of Univ. Our focus –Research, commercialization, education, collaboration
3 About NICTA 5 research labs –located in Sydney, Canberra and Queensland 14 Research programs –Empirical Software Engineering;Empirical Software Engineering –Interfaces, Machines, And Graphic EnvironmentsInterfaces, Machines, And Graphic Environments –Networks and Pervasive Computing.Networks and Pervasive Computing –Embedded, Real-Time, and Operating SystemsEmbedded, Real-Time, and Operating Systems –Formal MethodsFormal Methods –Symbolic Machine Learning and Knowledge AcquisitionSymbolic Machine Learning and Knowledge Acquisition –Statistical Machine Learning;Statistical Machine Learning –Systems Engineering and Complex Systems Systems Engineering and Complex Systems –Wireless Signal ProcessingWireless Signal Processing –Logic and Computation;Logic and Computation –Autonomous Systems and Sensing TechnologiesAutonomous Systems and Sensing Technologies –Statistical Machine Learning.Statistical Machine Learning –Sensor Networks;Sensor Networks –Network Information Processing.Network Information Processing
VTC2005 fall4 Agenda Motivation –Search of the environmental factors that can be utilized for outage prediction for on-board communication Measurement results Conclusion and future work
VTC2005 fall5 Mobile Network Growing interest in providing broadband service for public transport passengers Mobile Network –On-board LAN –Mobile Router (MR) Standardized protocol (RFC3963): NEMO –Extension of MIPv6
VTC2005 fall6 Exciting on-board applications Data Server Mobile Router On-Board LAN entertainment VoIP surveillance
7 Products/services already appearing! Cisco mobile router
VTC2005 fall8 On the research side IETF NEMO working group [2002] Nautilus6 working group [Japan, 2003] Network Mobility project [Korea, 2003] WirelessCabin [Europe, 2002] Network On Wheels (NOW) [Europe, 2005] PATH project [Berkeley] Diesel project [Umass, Amherst]
9 Outages outage
VTC2005 fall10 Outage prediction Important problem for on-board communication –Can impact a large number of users Mobility pattern of public transport vehicles are known in advance –Utilize this feature for outage prediction –Mobile Router records signal strength and available bandwidth information at different times and at different locations –Predict outage by analyzing recorded information
VTC2005 fall11 Effect of environmental factors Various factors might affect quality of wireless signal While physical factors like noise, multi-path will obviously affect the signal strength, –In practice, not easy to utilize them for outage prediction Environmental factors such as location, weather, time of the day, crowdedness, vehicle velocity, etc. –easier to observe/obtain and use
VTC2005 fall12 Contribution of this work We conducted wide-area measurements by recording GPRS signal in different locations and under a variety of conditions in Sydney metropolitan area We found location is the most dominating environmental factor affecting signal quality Some cellular operators might have done this –but their measurements are not public available
VTC2005 fall13 Agenda Motivation Measurement results –preliminary results on location, speed, humidity, people and tunnel Conclusion and future work
VTC2005 fall14 Data collection We recorded signal strength of Vodafone’s GPRS network under different conditions in a 6-month period Measurements were taken on the link between the receiver and base station GPS is used to record location and time of measurements Traces in collected on train, bus, car and some chosen locations
VTC2005 fall15 location Signal strength is strongly correlated with locations across different times of the day
VTC2005 fall16 Signal strength map 3 bus routes Good quality: green Bad quality: red Similar patterns across different times and different days
17 Speed No significance in signal strength level Larger variations at a lower speed –Hypothesis: other environmental factors have a better chance to affect the signal at the lower speed More handoffs at a lower speed
18 Handoff sequences Sequence of switching between different base stations exhibit certain predictability Suggest the feasibility of deploying some resource reservation scheme for on-board network
19 Humidity Rainy: 2 millimeters per hour No significant difference Rain attenuation has a stronger in higher frequency band such as microwave link –GPRS uses 900MHz/1.8G –Some base stations uses microwave links to connect to Mobile Service switching Center (MSC)
VTC2005 fall20 People Crowded: –lunch time Non-crowded –Eastern break No significant differences Larger variation for crowded scenario
VTC2005 fall21 Tunnel Tunnel does not lead to continuous outages Micro-cell around the platform
VTC2005 fall22 Distribution of signal strength level Gaussian distributed
VTC2005 fall23 Throughput during the day Similar patterns across different days –Lower throughput during the business hours and in the evening Hypothesis: user behavior
VTC2005 fall24 Conclusion and future work We conducted wide-area GPRS measurements under a variety of conditions in Sydney metropolitan area We found location is a better predictor for outage prediction than other environmental factors Use measurement results to design practical outage prediction algorithm for on-board communication