sapfanet: spatially adaptive positioning for fanets
TRANSCRIPT
SAPFANET: Spatially Adaptive Positioning for FANETsJoshua Rentrope and Dr. Mustafa Ilhan AkbasComputer ScienceFlorida Polytechnic University
Design and Analysis of Intelligent Vehicular Networks and Applications (DIVANET)November 21, 2017
Table of Contents
1. Introduction2. Problem Definition3. Previous System4. Optimizations5. Simulation Results6. Conclusions
Introduction
● UAV and UAV Swarm technologies have advanced● Increased capabilities for UAVs
○ Better optical and auditory sensors○ More reliable communications○ Flexible controller interfaces
● Current research is effective swarm applications and autonomy
Problem Requirements
● UAV Swarm that can fill irregular environments (Odd spatial domains, Obstacles in environment)
● A formulation that can adjust UAV behavior to its environment● Applications of this protocol to hazardous tasks (search and
rescue, surveillance)
Problem Specification
● Already existing 3d Swarms do not give enough weight to the environmental constraints
● There may need to be a centralized system
● Objective○ Create optimizations that consider the environment and can dynamically
reposition UAVs into better configurations
Previous Systems
● Virtual Forces○ Adjusts drone positions based on nearby drones and their relations○ Good in theory, but it is expensive in terms of communication○ This approach also ignores environmental macro-behaviours
● VSEPR Drone Positioning○ Calculates the equilibrium for Virtual Forces based on the geometry of the atoms○ Optimizes the communication needed by the swarm by updating key positions
and pair numbers
VSEPR Positioning
● VSEPR provides the optimal positions from Virtual Forces
● Can be a hierarchical organization and decentralized from a main controller
● VSEPR does not take into account the environment. Less optimal as Environment changes
APAWSAN: VSEPR-based Positioning
Actor positioning for aerial wireless sensor and actor networks
● APAWSAN uses precalculated VSEPR positions for drone configurations● A leader drone will broadcast its position to the follower drones● Each follower drone will move to a its labeled position based on the formula
This works great for unbounded spaces, each drone will have optimal angles. However, this solution is not generalized for corridor-like spaces and with obstacles
SAPFANET: An Optimization
● Spatially Adaptive Positioning for FANETs merges the spatial constraints of the environment to generate drone positions
● SAPFANET assigns spatial sub-domains to drone positions and attempts to equalize them as much as possible.
● First it calculates the optimal subdomains.○ Each subdomain is defined by a set
boundary planes. ○ Some boundary planes are the
constraining planes of the environment○ Boundary planes we can manipulate are
planes that part between two drones○ A configuration, which keeps each volume
of the set as close to the average as possible, is estimated
SAPFANET
SAPFANET Results
When compared to APAWSAN, the pure VSEPR configuration, SAPFANET displayed favorable results in terms of space
The time of convergence was slightly higher, but can be reduced through better communication protocols
Conclusion
● Can adapt VSEPR to dynamically constrained environments● SAPFANET is mainly suited for static environment, and needs further adaptation
for high mobility systems● Future work
○ More tests with obstacles, discontinuities in environment○ Using more realistic antenna models for communication○ Currently designing a simulation for Search and Rescue using SAPFANET
configurations and belief networks