spatial locality in pocket switched networks
TRANSCRIPT
Spatial Locality inPocket Switched Networks
June 21st, 2016
Tekenate E. Amah*, Maznah Kamat*, Kamalrulnizam Abu Bakar*, Waldir Moreira†,¥, Antonio Oliveira-Jr‡, and Marcos A. Batista§
The 10th IEEE WoWMoM Workshop on Autonomic and Opportunistic Communications (AOC 2016) Coimbra, Portugal
*Dept. of Computer Science, Faculty of Computing, Universiti Teknologi Malaysia, Skudai, Johor, Malaysia | †COPELABS, University Lusófona, Lisboa, Portugal | ¥PPGMO, Federal University of Goiás, Catalão, Brazil | ‡INF, Federal University of Goiás, Goiânia, Brazil | §IBiotec, Federal University of Goiás, Catalão, Brazil
Agenda
Introduction
Motivation
Methodology
Results
Towards locality aware PSN routing
Conclusions and Future work
Introduction
Ubiquitous portable handheld devices Portable handheld devices form the center of human communication today
1
Store Carry Forward (SCF) communication Communication via short-range wireless interfaces, user movement and opportunistic encounters
2
Pocket Switched Networks (PSNs) = Handheld devices + SCF Opportunistic networks formed by user devices for non-real-time communications
3
Presents a scalable and resource efficient solution for data transfer Urban: connecting major infrastructure | Developing regions: providing non-real-time connectivity
4
Motivation
User movement exhibits spatial locality Users show preference to certain geographical locations or districts more than others
1
Locality affects encounter and data transfer opportunities Lower locality presents more encounter opportunities, and vice versa
2
PSN routing strategies depend on encounter opportunities There is need to understand the impact of locality on routing performance in PSNs
3
The impact of locality on point-to-point communication in PSNs How does locality affect PSN routing and how can such routing benefit from locality awareness?
4
Methodology The Helsinki city area in the Opportunistic Network Environment (ONE) Simulator 3 scenarios: locality is increased by including more overlapping districts
Small scenario Low locality: 1 district
Medium scenario Medium locality: 4 districts
Large scenario High locality: 16 districts
Methodology
Simulation parameters
Parameter
Total simulation time (days)
Warm up period (days)
Wireless interface
Transmission range (m)
Transmission rate (Mbps)
Buffer size (Mb)
Value(s)
5
1
Bluetooth
10
2
100
Message size (Mb)
Message TTL (days)
Number of nodes
Ave. meg generation/node
0.5 to 1
1
19, 76 & 304
1 msg/hr
Results
Average delivery ratio
• Delivery ratio decrease • Unlimited replication increases
buffer overflow • Lower delivery ratio for destinations farther away
• Spraying mitigates the ‘slow start’ phase
Average overhead ratio
• Overhead ratio increase • Increased locality requires more
replicas to reach destinations • Unlimited replication increases
overhead • Spraying controls the number of
replicas
Average delivery delay
• Delay increase (except SnF) • Transitivity in the ‘focus’ phase
(in SnF) reduces delay • Transitivity incurs storage
overhead
Towards Locality-aware PSN Routing
PRoPHET-LA
• Highly active nodes forward messages traversing multiple
regions • Others forward messages within
their home region • PRoPHET-LA decreases average overhead by 57%
SnF-LA
• The optimal value of L varies over different degrees of locality • The value of L for a message
depends on the home proximity, p, of the source-destination pair • SnF-LA varies L according to
L = 2(p+1)
PRoPHET-LA2
• PRoPHET-LA2 nodes maintain only encounter information for neighbours from their home
region • 11% less delivery ratio, 84%
less overhead and 5% less delay, with 88% less encounter
information stored
Conclusions and Future Work
Performance degradation
PSN routing performance decreases over higher locality
Storage overhead
Storage overhead increases with network size
Locality awareness
Locality awareness complements existing routing strategies
Next steps
A measure for locality
A study of the concepts of locality
Validation of the synthetic movement scenarios
Effects of locality on routing fairness and energy consumption