This document proposes using swarming drones equipped with network testing devices to load cellular networks and test performance. It describes using particle swarm optimization to coordinate the drones to efficiently scan for problematic network regions. The drones act as particles in the PSO algorithm, communicating to optimize their search. Both single and multi-objective PSO are suggested to detect areas with issues like low signal quality. This swarming drone approach could replace manual drive testing which is time-consuming and limited to accessible areas.
Loader and Tester Swarming Drones for Cellular Phone Network Loading and Field Test: Non-stochastic Particle Swarm Optimization
1. Cellular Phone Loaded Network Field Test Using
Swarming Drones: Replacing Drive Test and
Particle Swarm Optimization
Amir Mirzaeinia, Mostafa Hassanalian, Mehdi Mirzaeinia
2019 AIAAAVIATION Forum
17-21 June 2019
Dallas, Texas
New Mexico Tech
Mechanical Engineering Department
New Mexico Tech
Department of Computer Science and Engineering
2. 2
Outline
Background & motivation
Cellular Networks (2G, 3G, 4G, 5G)
Loaded Network (events)
Field Performance Test (Coverage, Throughput)
Equipped drones to load the network and test
Designed quadcopter
Designed Micro flapping wing minimum system
Methodology for loading and testing the RF
Particle Swarm Optimization (PSO)
How PSO works
Simulations and results
Multi objective optimization
4. 4
Background & motivation
[Open domain images]
Loaded Network (events)
New Year’s Eve
Super bowl Stadium
Great theater
Other international event
5. 5
Background & motivation
[Open domain images]
Field Performance Drive Test (Coverage, Throughput)
GPS, a laptop, and a phone setup is needed
6. 6
Background & motivation
Hard to perform: Drive test method is very time consuming in
particular for testing inside large buildings, such as theaters,
stadiums, or congested streets.
Setup time: This vehicle involving method also needs to use
numerous setup devices, to schedule the tester experts.
Other coordination: police department Coordination in order to
be able to drive around the city with all testing tools installed on it.
Inaccessible places : mountainous areas for skiers, national park,
inaccessible IoT sensors and actuators.
Field Performance Drive Test problem.
7. 7
Equipped drones to load the network and RF test
[Open domain images]
Equipped drones with network test handset to load the network and RF test ?
https://appliedmechanicsreviews.asmedigitalcollection.asme.org/article.aspx?articleid=2667680
Designed Micro flapping wing minimum system
8. 8
Equipped drones to load the network and test
Proposed methodology for loading and testing the network
11. 11
Particle Swarm Optimization (PSO)
Particle Swarm Optimization (PSO)
𝑋𝑖=(𝑥𝑖_𝐿𝑎𝑡,𝑥𝑖_𝐿𝑜𝑛𝑔)
𝑃𝑖=(𝑝𝑖_𝐿𝑎𝑡, 𝑝𝑖_𝐿𝑜𝑛𝑔)
𝑉𝑖=(𝑣𝑖_𝐿𝑎𝑡,𝑣𝑖_𝐿𝑜𝑛𝑔)
𝑉𝑖𝑑=𝑉𝑖𝑑+𝑐1×𝑟𝑎𝑛𝑑()×(𝑝𝑖1−𝑥𝑖2)+ 𝑐2×𝑟𝑎𝑛𝑑()×(𝑝 𝑏𝑑−𝑥 𝑏𝑑)
𝑥𝑖𝑑=𝑥𝑖𝑑+𝑣i𝑑
𝑉𝑖𝑑=𝑤×𝑉𝑖𝑑+𝑐1×𝑟𝑎𝑛𝑑()×(𝑝𝑖1−𝑥𝑖2)+ 𝑐2×𝑟𝑎𝑛𝑑()×(𝑝 𝑏𝑑−𝑥 𝑏𝑑)
Location of the ith drone
Best position of ith particle
Velocity and location update in each iteration, b represents the best particle
Velocity of the particle
Inertia weight is used in PSO to find a balance between global and local search
15. 15
Particle Swarm Optimization (PSO)
Multi objective PSO deploying in swarming drones
External interferences are affecting on service quality
High level signal but low quality
PSO can also detect these areas
min(−𝑅𝑥𝐿𝑒𝑣𝑒𝑙 ,𝑅𝑥𝑞𝑢𝑎𝑙𝑖𝑡𝑦)
16. Conclusions
19
Network Loading: Using swarming drones to load a cellular network and test the
performance is very desired to have more realistic testing situations.
Manual Navigation: Swarming drones manual navigation to scan the network
performance and find the performance problematic region is time-consuming and
not very efficient to deploy.
Swarming Drone Optimization: inspired by particle swarm optimization,
swarming drone optimization is proposed. In proposed method each drone works
as particle agent. Swarming drone communication to deploy the PSO in all drones
helps to find the network problematic regions.
Multi-objective optimization: Besides, there are cases in cell phone network
optimization which need multi-objective optimization. Hence, Multi-objective PSO
optimization can be applied in swarming drone to find more complicated
problematic regions.
17. Thank you for your attention
Questions?
amir.mirzaeinia@student.nmt.edu
mostafa.hassanalian@nmt.edu