Tutorial on Wireless Communications and Networking with Drones and Unmanned Aerial Vehicles (UAVs)

Wireless Communications and Networking
with Unmanned Aerial Vehicles (UAVs)
Tutorial
Walid Saad
Wireless@VT, Electrical and
Computer Engineering Department,
Virginia Tech, USA
walids@vt.edu
http://resume.walid-saad.com
http://www.netsciwis.com
Mehdi Bennis
Center for Wireless
Communications
University of Oulu, Finland
bennis@ee.oulu.fi
http://www.ee.oulu.fi/~bennis/

Outline
 Introduction and motivation
 Part I: Channel modeling for UAVs
 Part II: Performance analysis and tradeoffs
 Part III: Optimal deployment
 Part IV: Resource management for UAVs
 Part V: Security
 Concluding remarks
2

The inevitable rise of the UAV
3
Few facts:
 The number of UAVs will skyrocket
from few hundreds in 2015 to
230,000 in 2035
 Different types of aerial
objects/systems, LOS, BLOS
 Includes drones, LAP, HAP,
balloons, quadcopters, etc
Facebook Project AquilaGoogle Project LOON
OneWeb LEO constellation: 648 low-
weight, low orbit and low latency
satellites positioned around 750 miles
above Earth …+ SpaceX from E. Musk
Matternet

 Can be a small plane, balloon or drone
 High altitude platform (HAP) above 15 km, or Low altitude platform
(LAP) between 200 m to 6 km
 Proposals from Facebook, Google, spaceX to connect the unconnected
 Frequency bands for HAPs: 38-39.5GHz (global), 21.4-22 GHz
and 24.25-27.5GHz (region-specific)
 Remotely controlled or pre-programmed flight path
 Control and non-payload communication (CNPC) systems
4
Unmanned Aerial Vehicles

5
Deutsche Telekom and Inmarsat partner to deliver European
Aviation Network w/ a target low latency of 40ms.
• Beam switching between
ground-BSs and Satellites
• Other telcos following same
trend
• Satellite community meets
telcos
Boosting Connectivity in the Sky

 Applications
 Communications, disaster management, search and rescue,
security, control, agriculture, IoT, etc
 Covering hotspots
+ 1000x more
 Advantages
 Adjustable altitude
 Potential Mobility
 Low infrastructure low cost
 Limited available energy for Drones
 Also, many challenges
6
Countless Applications
disaster
Coverage/capacity
V2V
Smarter
mobility
VR
Agriculture

7
Challenges
Deployment
Path planning/mobility
Energy efficiency
Channel modeling
Interference
Handover and
moving cells
Security and privacy
Resource
management

Wireless Back-/Fronthauling
 UAV-to-UAV communication required for coordination, interference
mitigation, relaying, routing in the air, etc.
 Satellite and WiFi considered as candidate technologies for providing
wireless backhauling depending on latency-bandwidth requirements
 Satellite backhauling brings the advantage of unlimited coverage
offering the possibility of connecting the aerial network for any
distance
 However, the latency introduced by the satellite links (GEO) may affect
some real time services such as voice and real-time video.
 To avoid satellite delays and the cost, WiFi links can be used albeit
reduced coverage and capacity (doubtful QoS guarantees..)
Recent interest in Free Space Optics
 License free PtP narrow beams
 But tackle rain, fog and cloud attenuations
 Multi-connectivity to the rescue..? 8
Backhaul

9
Tools Usefuls for UAVs
5G+5G+
Physics
• Mean field
• Random graphEconomics
• Matching theory
• Pricing
Game theory (GT) and learning
• Decision making
• Resource management
• Clustering
• Supervised, non-supervised learning
Control Theory
• Lyapunov
• Consensus
Stochastic geometry
• BS/UE location
Stochastic optimization
• CSI/QSI uncertainties
Random matrix theory
• Asymptotics
Transport Theory
• Association
• Mobility
In this tutorial, we will (briefly) touch on
GT, optimal transport, and learning

10
Part I –
Air-to Ground
Channel Modeling
for UAVs

11
Air-to-Ground AtG Channel Model
 Radio propagation in AtG channel differs from terrestrial
propagation models
 Typically radio waves in AtG channel travel freely without obstacles
for large distances before reaching the urban layer of man-made
structures.
 UAV-ground channels typically include:
 Line-of-sight (LOS) and NLOS links
 A number of multi-path components (MPC) due to reflection, scattering, and
diffraction by mountains, ground surface, foliage
 Common models define a LOS probability
between UAV and ground user that depends on:
 Environment (suburban, urban, dense urban)
 Height (h) and density of the buildings (building/km2)

12
Air-to-Ground Channel Model
 Received signals include:
 Line of sight (LOS): strong signal (G1)
 Non-line of sight (NLOS): strong reflection (G2) or fading (G3)
 Each group with a specific probability and excessive loss
 Dominant components
 LOS links exist with probability P and NLOS links exist with
probability 1-P
 Consider LOS/NLOS separately with different path loss values
Excessive path loss
samples
histogram

13
Air-to-Ground Channel Model
 Model by Al-Hourani et al.
 Buildings and environment impact the propagation
 Distribution of buildings’ heights:
 Suburban
 Urban
 Dense urban
 Highrise urban
A scale parameter depends
on environment according
to a Rayleigh pdf

14
Ray Tracing Simulation
 Allows the prediction of signal strength in an accurate
manner
 Based on a simulation of actual physical wave
propagation process
 Can consider different ray types: Direct, Reflected and
Diffracted rays
 Requires buildings database
 3D predictions

15
Ray Tracing Simulation
 Propagation Group Occurrence Probability, obtained at
frequency = 2 GHz for an urban environment
 Group 1: LOS
 Group 2: NLOS
Example of a group occurrence curve ﬁtting for two groups
 Occurence probability
of a certain propagation
group at a certain angle

Parameter depends on
environment
Back to LOS Probability
 In urban environments, the LOS probability is given as:
16
ratio of built-up land area to the total land
area mean number of buildings per
unit area (buildings/km2 )


Antenna height
…For large values of h,P(LoS) is a
continuous function of θ and environment
parameters see next slide

17
LOS Probability approximation
 Probability of having LOS link:
 Trend approximated to a simple modiﬁed Sigmoid function (S-
curve)
 Increasing LOS probability
by increasing elevation angle or
UAV’s altitude
B and C: constants that depend
on the environment
θ: Elevation angle

18
Shadow Fading
 Modeling shadow fading
Received signal
power
Shadow fading
Gaussian
distribution
Parameters depend
on environment

19
Ricean channel model
 Small scale fading is described by the Rician distribution
due to the presence of a strong LOS component in the
AtG channel
 Distribution of the received signal amplitude:
 Rician K factor:
 Depends on the environment
 Lower for denser environments
LOS amplitudeAverage multipath
component power
Bessel function
L-BandRicean K-factors = 12 dB and 27.4 dB in
C-band in the near-urban environment.
14 dB in L-band and 28.5 dB in C-band for the
suburban settings.

Way forward
 Air-to-air channel models (still lacking in literature)
 The probabilistic model may not be the best, real-
world measurements can help
 Airframe shadowing for large-sized or small-sized
aircraft, tree/building shadowing at low altitude small
UAV, also terrain shadowing for mountainous
scenarios or beyond LOS conditions
 Of relevance here are the works of Matolak and NASA
 How to integrate multiple antennas, what is the most
adequate number of elements and their location
(MIMO or mmWave air-to-ground channels?) 20

21
Part II –
Performance
Analysis

22
Drone small cells in the clouds:
Initial insights on design and
performance analysis

System Model
 Downlink scenario
 Drones provide coverage for a target area
 Scenarios:
 Single drone
 2 drones without interference
 2 drones with intercell interference
 Target: Meeting the minimum SINR
requirement on the ground
23

Main Goals
 Determining the optimal altitude of drones
 Leading to maximum coverage
 Full coverage using minimum transmit power for the drones
 Optimal deployment of two interfering drones
 Distance between the drones?
 Altitudes?
 Highlighting tradeoffs while deploying drones
 Interference, coverage, transmit power
24

Impact of Drone’s Altitude
 Higher altitude: Higher path loss vs. higher LOS proba.
 Lower altitude: Lower path loss vs. more NLOS
 Altitude and flight constraints
 Higher and lower altitudes are bounded
25
What is
the Optimal
Altitude?

Single Drone
 Minimize transmit power via an optimal altitude
 Path loss as a function of elevation angle:
26
Environmental parameters
Additional loss for NLoS
Optimal altitude

Optimal Altitude
 Optimal altitude depends on the area size (Rc)
 Increasing drone’s altitude to service larger areas
27
@Low-altitude: high shadowing
+ low LOS probability coverage radius decreases
@ high-altitude: high LOS probability but PL
Increases –> Coverage decreases
E.g.; optimal altitude for providing 500m coverage
radius while consuming min. tx power is 310 meters
Altitude increases w/ coverage radius

Two Drones
 Given a desired geographical area:
 Maximize the total coverage area
 What is the distance between drones?
 What is their altitude?
28
Total coverageDistance between
drones

No Interference Case
 Deploying each drone at its optimal altitude
 Packing the coverage areas inside the target area
 While keeping the distance between drones as far as
possible, but inside the target area
29
Maximum coverage
range of each drone
Total coverage

Two Interfering Drones
 Consider a rectangular geographical area
 High distance between drones: covering undesired area
 Small distance between drones: high interference
30
• No coverage in between due to the interference
• Drones should not be placed too close
• 300 meter altitude
• 1100 meter separation

Results
 Bounded target geographical area
 Existence of optimal drones’ separation distance for
maximum coverage
31
At high drone distance, although separated,
coverage ratio is low (undesired)
Likewise, if too close interference increases.
optimal separation distance exists!

32
Unmanned Aerial Vehicle with
Underlaid Device-to-Device
Communications: Performance and
Tradeoffs

System Model
 Downlink Scenario: UAV coexists with a device-to-device (D2D)
network
 Two types of users: downlink users (DU) and D2D
 UAV provides service for downlink users
 Interference between UAV and D2D transmitters
 Static and Mobile UAV Cases
33

UAV and D2D: Assumptions
 Users (DU and D2D) distributed based on Poisson point
process (PPP)
 Number of users follows Poisson distribution, but uniformly
distributed over the area
 The number of points in a bounded area has a Poisson
distribution with mean e.g. λ×A or λ×B
 Underlay D2D communications:
use existing licensed spectrum
 Can we analyze the performance
tradeoffs for UAV deployment
34

 Derive the average coverage probability and sum-rate
expressions
 Finding the relationship between UAV parameters (altitude,
etc.) and rate/coverage
 Finding some fundamental performance tradeoffs
What is the optimal altitude of the UAV that maximizes
the coverage and rate?
 Fundamental tradeoffs between DU and D2D users
How to optimize coverage using UAV mobility ?
Main Objectives
35

 Coverage probability for downlink users (DUs)
 Coverage probability for D2D users
 Average rates
Performance Evaluation Metrics
36
SINRSINR ThresholdPolar coordinates

Static UAV: Analytical Results
 D2D Coverage Probability
 DU Coverage Probability
37Interference from D2D links
UAV
transmit
power
D2D
transmit
power
UAV-D2D
distance
Distance
Between
D2D pairs
D2D
density
LoS
probability

 Number of D2Ds
 Impacts interference generated at the DUs
 Distance between each D2D pair
 UAV’s location and altitude
 Impacts air-to-ground channel
 Transmit powers of D2D and UAV
 Directly affect the coverage probabilities
 SINR threshold
 Overall, we have a tractable expression to analyze UAV
coverage
Key parameters
38

Results: Static UAV
 Optimal altitude for DU maximum coverage
 LoS and NLoS tradeoff
 Impact of altitude on D2D coverage probability
 UAV is an interference source for D2D
39
Optimal

Results: Static UAV
 Average sum rate vs. altitude
 Considering DU and D2D rates
 Depends on the distance between each D2D pair
40
The lower is the higher
is the sum-rate

Mobile UAV
 UAV moves over the target area
 Transmits at given geographical locations: “stop points”
 Goal: satisfy DUs coverage requirements by covering
the entire area
 Analyze the impact of UAV’s mobility on the outage
probability of D2D links
 Considering the spatial correlation in D2D communications
Question: What is the minimum number of stop points
(delay)? 41

Mobile UAV
 Minimum number of stop points
 Depends on: UAV altitude, D2D density, size of area,
coverage constraint
 Moving the UAV to provide complete coverage
for the area of interest
 Using optimal circle covering approach
 Full coverage with minimum
number of circles
42

Results: Mobile UAV
 Maximum coverage radius vs D2D density
 Higher number of D2Ds: higher interference
 Decreasing coverage radius!
43

Results: Mobile UAV
 Number of stop points vs. D2D density
 Higher number of D2Ds: higher interference
 Increasing number of stop points!
44

Results: Mobile UAV
 Altitude and number of stop points
 : target DU coverage requirement
 Altitude impacts coverage range and thus number of stop
points
45
Higher coverage requires
more stopping points

Results: Mobile UAV
 Coverage-delay tradeoff
 Higher number of stop points:
 Better coverage performance for DUs
 Leads to a higher delay
46

Results: Mobile UAV
 UAV affects the D2D outage
 No UAV: only other D2Ds create interference
 With UAV: UAV+ other D2Ds are interference sources
 Moving UAVs leads to higher average outage probability for
D2D network
47

48
Part III – Optimal
Deployment

 Where and when to deploy UAV-BSs?
 What metrics to optimize (long term vs. short term)?
 How to develop wireless-aware path planning mechanisms?
Optimal Deployment and Mobility
49
UAV Base Stations (LAPs) Terrestrial Base Stations
• Deployment is three-
dimensional
• Deployment is two-
dimensional (with small
exceptions)
• Short-term, frequently
changing deployments
• Mostly long-term,
permanent deployments
• Mostly unrestricted locations • Few, select locations
• Mobility dimension • Fixed and static

50
Deployment strategies of multiple
UAVs for optimal wireless coverage

System Model
 Downlink communications
 Using directional antennas for UAVs
 Interference between all UAVs
 Circular target area
 Meeting the minimum SINR
requirement on the ground
51

 Derive the coverage probability and coverage range of
each UAV
 Maximize the coverage performance by efficient
deployment of multiple UAVs
 Adjust UAVs’ altitude based on antenna beamwidth
 Avoid overlapping coverage to avoid interference
Main Objectives
52

 Considering shadowing effect in LoS and NLoS links
,
Downlink Coverage Probability
53
Received signal power
Path loss
3 dB antenna gainQ function

 Coverage range of each UAV:
 M identical UAVs
 Total coverage is maximized
 No overlap between UAVs’ coverage areas
Multiple-UAVs deployment
54

 Big circle: area of interest which needs to be covered
 Each small circle: Coverage region of each UAV
Maximizing the packing density is equivalent to
maximizing total coverage
Approach: Circle Packing Problem
55
The optimal
packing of 10
circles in a circle
The optimal packing
of 15 circles in a
square
The optimal packing
of 6 circles in a
right isosceles
triangle

 Coverage radius vs. number of UAVs (circle packing):
 Upper bound on the coverage radius:
Results
56

 Altitude versus number of UAVs
 More UAVs:
 Less coverage radius per UAV is required
 Reduce UAVs’ altitudes to avoid interference (overlapping)
Results
57

 Meeting a total coverage requirement
 What is the minimum number of UAVs?
 Depends on the size of the area
 Choosing appropriate number of UAVs based on coverage
requirement and size of target area
Results
58

 Total coverage and coverage lifetime tradeoff
 Increasing number of UAVs:
 Transmit power per UAV can be reduced
 Higher coverage lifetime
Results
59

60
Cooperative deployment and mobility
of UAVs for optimizing rate-delay
tradeoffs

Cooperative UAV Deployment
Task 3Task 2 Task 1
Task 4
 Given a number of tasks in an area and some autonomous
agents (e.g., UAVs)
 How to dispatch the agents to service the tasks?
 Can the agents make their own decisions on servicing the tasks?
 Almost no work considered the problem in the context of a
wireless/communication network
 Tasks are queues of data with no direct connectivity
Task 6Task 5
61

62
 The problem is well studied but…
 Most approaches are
 Robotics-oriented
 Mainly in military applications (tasks are targets)
 Other related problems (the repairman problem, dynamic vehicle
problem… )
 Software engineering (autonomous agents)
 The tasks are usually considered as passive entities
 Almost no work considered the problem in the context of a
wireless/communication network
 With next generation self-organizing networks this problem becomes quite
relevant
 Nature of wireless networks (channel, traffic, etc)
 Quality of service
Cooperative UAV Deployment

63
Agents in Wireless Networks
 Given a number of tasks in an area
 Consider each task as a M/D/1 queuing system generating
packets with a Poisson arrival
 Each task i has an arrival rate λi
 The network operator, requires..
 Data collection from the tasks
 Wireless transmission of the data to a central receiver
 The network owns a number of autonomous agents
that need to
 Decide on which tasks to service
 Collect the data and transmit it taking into account
 The amount of data collected
 The delay

Task 4
Task 3
Task 5
Task 2
UAV Agents in Wireless Networks
Central Receiver
(Command Center)
Task 1
 How will such groups form?
 A cooperative game between Tasks and Agents
 Solution using notions from operations research, wireless
networks, and queuing theory
Collect data from task
5 and transmit it
3 and transmit it
4 and transmit it
Collect Data from
Task 1 and transmit it
Collect Data from
Task 2 and transmit it
64

65
Problem Formulation
 Coalitional game where
 The players are the tasks and UAVs, hence, the player set N is the
set of all tasks and UAVs
 Denote M the set of UAVs and T the set of tasks, N = M U T
 Each coalition S consists of a number UAVs servicing a number of
tasks t
 A UAV can be either
 A collector: more collectors means smaller service time, less
delay
 Each collector i has a link transmission capacity µi
 For a number of collectors G servicing a task i in a coalition S the
total link transmission capacity is
 A relay: more relay means better effective throughput (less outage
probability)

66
Problem Statement
 Each coalition S can be seen as a polling system
with exhaustive strategy and switchover times
 Polling systems are ubiquitous in computer systems
 The main idea is that a server is servicing multiple
queues (sequentially or not)
 Exhaustive implies the server collects all the available data
from a queue before moving to the next
 Switchover times are the time to move from one task to
the other
 In this context, each coalition S consists of
 A number of collectors acting as the polling system
server
 The tasks are the queues of the polling system
 Switchover time is the travel time from one task to the
next

67
Performance metrics - Delay
 For a polling system, it is difficult to have an exact
expression for delays, but, we can use the pseudo-
conservation law for a coalition S
 Stability of coalition S (polling system) requires ρS < 1
 The total switchover time θS depends on the sequence in
which the tasks are visited
 Nearest neighbor solution to the travelling salesman problem
Utilization factor ρi:
ratio of arrival rate to link
transmission capacity of
collectors for a task i
Total switchover time
θS of coalition S
Sum of utilization
factors over all tasks
in S

68
Performance metrics - Throughput
 For each coalition, the total effective throughput
from the data collected and transmitted is given by
 Pri,BS is the outage probability for wireless transmission
from task i to the central BS
 Improved by having UAVs working as relays on the link
between the collectors on task i and the BS
 For each coalition, the UAVs and tasks are given a
reward from the network operator depending on
the throughput-delay trade off achieved

69
Utility function
 Given the throughput and delay previously defined, for
each coalition S we propose the following utility
 β is a tradeoff parameter that represents the weight that a coalition
puts on the throughtput and delay
 The utility is based on the concept of power which is a ratio
between effective throughput and delay
 Utility is transferable: the total revenue achieved by coalition S
with δ the revenue per unit power
 Given the players set N and the utility v the question is
 We use the framework of hedonic coalition formation games to
solve the problem

70
Hedonic Coalition Formation
 In our game we can say that
 A UAV prefers a coalition S1 over a coalition S2 if
 The UAV is not the only UAV in its current coalition S2 and
 The payoff he receives in S1 is higher than S2, and he had not
visited this coalition before (history tracking).
 xi
S is the payoff received by player i from the division of the utility
(we consider equal division for this work)
 h(i) history set

71
Hedonic Coalition Formation
 A task prefers a coalition S1 over a coalition S2 if
 The payoff he receives in S1 is higher than S2, and he had not
visited this coalition before (history tracking).
 By using these preferences we can derive an algorithm
form coalitions between the UAVs and the tasks
 Having defined the preferences, the next question is
 How to form the coalitions?

72
Coalition Formation Algorithm
 Coalitions form and break as a result of
selfish decisions by the players (agents and
tasks)
 Switch rule
 Every player switches its current coalition to
join another, if and only if the new coalition is
strictly preferred using the defined preferences.

73
Coalition formation algorithm
Initial Network State:
Non-cooperative network
Task discovery:
The central BS informs the UAVs
of the tasks in the area
Each player ( UAV or task) surveys
nearby coalitions for possible switch
Each player takes the switch
Decision that maximizes its payoff
Sequential switch operations until convergence
Final partition: Continuous data
collection and transmission by the UAVs
The final
partition is
Nash-stable,
no player has
an incentive to
unilaterally
change its
coalition

76
Mobile UAVs for Energy-Efficient
Internet of Things Communications

System Model
 Uplink IoT communications
 Meeting SINR requirements of IoT devices
 Periodic versus Probabilistic IoT activation models
 UAVs update their locations based on devices activation
patterns
77

IoT devices
 Battery limited
 Typically unable to transmit over a long distance due to their
energy constraints
 UAVs can dynamically move towards IoT devices, collect the
IoT data  moving IoT aggregators
 Many IoT devicesinterference issue
 IoT activations:
 Periodic: weather monitoring and smart grids applications
 Probabilistic: health monitoring and smart traffic control
applications.
78

 How to enable reliable and energy-efficient uplink
communications in a large-scale IoT using UAVs?
 What are the joint optimal 3D UAVs’ locations, device-
UAV associations and uplink power control?
 Need for a framework for updating UAVs locations in
time-varying networks:
1) Update times: shows how frequently UAVs update their
locations
2) UAV trajectories
Main Objectives
79

 Joint UAVs’ locations, associations, and power
optimization
Problem Formulation
80
IoT transmit
power
UAV j
location
Set of active
IoT devices
SINR
Constraint
Channel
gain
Association
matrix

 Decompose the problem into two subproblems
 Solve the problem for fixed association
 Solve the problem for fixed UAVs’ locations
 Consider interference and non-interference scenarios separately
General Approach
81

 UAVs’ locations and device-UAV association
 An example, given the locations of active IoT devices
Results
82
 5 UAVs serving 100 active IoT devices uniformly distributed over the area

 Reliability
 Probability that active devices are successfully served by
UAVs
 Significant enhancement by dynamically moving UAVs
Results
83

 Total transmit power vs. number of UAVs
 Compared with stationary aerial base stations
Results
84
• 5% power reduction
vs. baseline on the
average

 Total transmit power vs. number of orthogonal channels
for meeting SINR requirements
 More channels:
 less interference and hence, lower transmit power needed to meet
SINR requirements of each device
Results
85
• 100 devices served by 5 UAVs
• By increasing the number of channels
from 25 to 50, the total transmit power
of devices can be reduced by 68%

 Time varying IoT network
 UAVs dynamically update their locations based on IoT
activations
 Probabilistic activation during [0,T]:
 Beta distribution with parameters
 Periodic activation:
 Each device has a specific activation period
IoT activation models
86

 Time instance at which the UAVs’ locations and
associations are updated
 Depends on the activation process of IoT devices
 Number of IoT devices
 For higher number of devices more updates are needed!
 Energy of the UAVs
 More updates requires more mobility
UAVs’ update times
87

 For probabilistic activation case
 Choosing appropriate update times based on
number of active devices
UAVs’ update times
88
Regularized incomplete beta
function
Average number of
active devices
Total number of
devices

 Number of devices which must be served vs. update time
 More frequent updates:
 More devices can be served
 Less active (unserved) devices remain
Results
89
For a higher number of update times or
equivalently shorter time period between
consecutive updates, the average number
of devices that need to transmit their data
decreases

 Update times for different number of active devices
 Depends on the activation process (beta distribution parameters)
 Ensuring that the average number of active devices is less than a
Results
90
Average number of
active devices
• To achieve lower value of a, updates need to be done more frequently so as the time
interval between updates decreases.
• For e.g., to meet a = 100, 75, and 50, the 5th update must occur at t = 0.41, 0.55, and 1

 UAVs update their locations according to the activity of
the IoT devices
 How to optimally move UAVs between the initial and
the new sets of locations?
 Mobility with minimum total energy consumption
 Energy consumption of each UAV depends on travel distance,
UAV’s speed and power consumption as function of speed
UAVs’ mobility
91
Travel time

 Which UAV goes where?
UAVs’ mobility
92
Update time t1Update time t2
New set of UAVs’
locations
Initial set of UAVs’
locations
Energy constraint
of each UAV
Can be
transformed
into an
assignment
problem
Transportation matrix
Energy from location k to l

 Update times impact the UAVs’ energy consumption for
mobility
 More updatesUAVs need to spend more energy on mobility
Results
93
• by increasing the number of updates from 3 to 6, the energy consumption of
UAVs increases by factor of 1.9

94
Part IV – Resource
Management

 Let’s first take a look on the impact of hover time
Resource management
95
UAV Networks Terrestrial Networks
• Spectrum is scarce • Spectrum is scarce
• Inherent ability for LoS
communication can facilitate
high-frequency (mmW)
• Difficulty to maintain
LoS poses challenges at
high frequencies
• Elaborate and stringent
energy constraints and models
• Well-defined energy
constraints and models
• Varying cell association • Static association
• Hover and flight time
constraints
• No timing constraints, BS
always there

96
Optimal Transport Theory for Hover
Time Optimization

Flight Time Constraints?
 UAVs have limited on-board batteries
 Cannot fly for a long time
 Flight regulations and weather conditions
 No-fly time and no-fly zones
 Wind and rain effects
 Mobility based on demands
 Cannot stay at one location for a long time
 Flight time constraints must be taken into account:
 Minimizing flight time while meeting the demands
 Optimizing the service performance under flight time constraints
97

System Model
 M stationary UAVs serve N users
 Users’ distribution:
 2D spatial distribution of users
 Determines how likely a user is present
 M partitions each serviced by one UAV
 Hover time: Time duration that a UAV spends over a given area
 Channel model adopted is the one explained earlire
 Goal: finding optimal cell partitions and associations
 Based on users’ distribution, hover times, and UAVs’ locations
 Two scenarios:
 Maximizing total service data given the maximum hover times (Scenario 1)
 Minimizing average hover time while meeting load requirements (Scenario 2)
98

Problem Formulation (Scenario 1)
 Total bandwidth for UAV i :
 Hover time of UAV i :
 Effective data transmission time:
 Control time which is not used for transmission:
 Portion of hover time which is not used for data transmission
 Used for processing, computations, and control signaling.
 Is a function of the average number of users
 Data transmitted to a user located at (x,y) served by UAV i :
99

Scenario 1
 Time and bandwidth are the resources
 We consider some level of fairness in resource allocation:
 Maximizing average total data service by optimal partitioning:
100
Depends on hover times and
bandwidths

Approach: Optimal Transport Theory
 Moving items from a source to destination with minimum cost
 What is the best way to move piles of sand to ﬁll up given holes of
the same total volume?
 Goal: Minimizing total transportation costs
 Where should each pile be moved?
 Our problem: transportation from users to UAVs!
101

Back to our problem
 We have a semi-discrete optimal transport problem
 Mapping from users’ distribution (continuous) to UAVs (discrete)
 Optimal cell partitions are related to optimal transport maps
102

Finding Optimal Partitions and
Associations
 Finding optimal values of leads to the optimal transport map
and optimal cell partitions!
 Complete characterization of partitions is now possible
103
Kantorovich Duality Theorem:
Theorem 1:
Cost function depending
on data service

Finding Optimal Partitions and
Associations
1) F is a concave function of
2) Using gradient based method to find optimal
3) Optimal cell partitions are given by:
104
Theorem 2:
Special case: results in a weighted Voronoi diagram

Results: Scenario 1
 We consider truncated Gaussian distribution for users
 Suitable for modeling hot spots in which users are congested
105

Results: Scenario 1
 Lower : users’ distribution is more non-uniform
 Jain’s fairness index is one when all users receive equal service
106
Average number of users in each
partition
Fairness index for average data service

Scenario 2
 UAV-based communications under load constraints
 Goal: minimizing the average hover
time needed for serving the users
 By finding optimal cell partitions
107
Average hover time of UAV i to
service partition :
Transmission time Control time
: rate
Load (in bits)

Problem Formulation (Scenario 2)
 Average total hover time of UAVs:
 We will characterize the optimal solution using optimal
transport theory again 108

Optimal Partitions
 Proof idea:
 Proving the existence of solution
 Comparing optimal partitions and a non-optimal variation of those
 Then characterizing the solution
 Note: weighted Voronoi is a special case (with no control time)
109
Theorem 3: optimal cell partitions can be characterized as

Results: Scenario 2
 Average hover time vs. control time
110

Results: Scenario 2
 Hover time and bandwidth tradeoff
111

112
Regret Based Learning for UAV assisted
LTE-U/WiFi Public Safety Networks

Learning-based UAV-assisted LTE-
U/WiFi Public Safety Networks
 Due to their mobility and self organization features,
UABSs are uniquely suited for delivering broadband
connectivity during public safety scenarios
 In addition to licensed spectrum, tap into unlicensed
bands
113

 Develop a game theoretic framework for load
balancing between LTE-U UABSs and WiFi access
points (APs), based on 1) users’s link qualities, and 2)
load of the UABSs and the APs
 We propose a regret-based learning (RBL) dynamic
duty cycle selection method for configuring
transmission gaps in LTE-U UABSs, so that a
satisfactory throughput is maintained for all users.
114

115
 Hybrid UABS-WiFi Architecture composed of:
 Ground based WiFi APs
 LTE-U UAVs at (given and fixed) optimal locations
 Transmission on the 5GHz frequency band with 20MHz bandwidth
 Focus on downlink
 Goal of LTE-U UABSs is to maximize their capacity while
keeping minimum interference towards WiFi transmissions.
Path loss model
• Air to ground from UABS to UE with LOS and
NLOS components
Traffic model
• traffic load of UE at location x follows an
inhomogeneous Poisson point process
and file sizes are independently
distributed with mean
• system-load density

 Ground-User association at location with BS
 Each BS will compute its estimated system load density as a
moving-time average as follows
 ON/OFF strategy of LTE-U UABS transmissions to coexist with
WiFi APs.
 UABSs autonomously and dynamically select their transmission
duty cycle (DDCS) in order to maximize their utility functions
 Mutual interference gives rise to a ”non-cooperative game” 116
Received SINR
learning rate

 Mutual interference gives rise to a ”non-cooperative game”
 set of actions:
 transmission duty cycle {0.2, 0.4, 0.6, 0.8}
Learning-based UAV-assisted
LTEU/WiFi Public Safety Networks
Utility of player iStrategy space
LTE-U frame representation for a duty cycle period of 0.6
 Average cell throughput of the i-th UABS for one duty cycle (50ms)
number of allocated resource blocks

 Our goal is to develop a distributed mechanism to
solve the LTE-U ON/OFF game and reach the
correlated equilibrium (CE)
 Since the mixed strategy Nash equilibrium (NE) tends
to have low efficiency CE is chosen over NE
 A class of no-regret learning algorithm called regret-
matching which converges to the CE

 Each layer consists of 7 cells.
 10 WiFi users and 10 LTE-U
users, associated with each WiFi
AP and LTE-U UABS
 Traffic arrival rates are given
119
Aggregate WiFi and LTE-U DL capacity for different duty cycles
(h = 50 m, ISD = 50 m).

120
 With fixed LTE-U duty cycle of 0.2,
SINR values of WiFi users are greater
than DDCS based procedures, due to
lower LTE-U air time
 WiFi SINR improves at larger heights due
to lower interference from LTE-U UABSs
 larger UABS heights will also result in
lower LTE-U capacity

121
 RBL-DDCS capacity of LTE-U downlink, WiFi downlink, and aggregate LTE-
U and WiFi transmissions for different ISDs and different traffic arrival rates
 LTE-U UABS capacity is significantly degraded for larger ISDs, while a
relatively lower degradation is observed for WiFi

122
Caching in the Sky: Proactive Deployment
of Cache-Enabled Unmanned Aerial
Vehicles for Optimized Quality-of-
Experience

System Model
 Considerations:
 Users mobility
 Users’ content request
 Caching at UAVs
 UAVs’ deployment
 Transmission links
 (a) Content server->BBUs->RRHs->users
 (b) Content server->BBUs->UAV>users
 (c) Cache->UAVs->users 123

 Maximizing users’ quality of experience (QoE) using
minimum UAVs’ transmit power
 Optimizing
 Users association
 UAVs’ locations
 Content caching
Main Objectives
124

General Approach
 For learning and predictions, we use the neural network
framework of echo state networks 125

 Notion of “reservoir” (random)
 Only need to train the output layer
via linear regression
 Good at dealing with time stamped data
Echo State Networks
Win
W
x n 
y n 
W
y n1 y n 
Win W
Wout
Wout
Wout
Wout
x n1 x n 
W
Win
Win W
y n2 
x n2 
u n  u n  u n1  u n2 

Echo State Networks
Step 3.
Training Process
Step 1.
Step 2.
Usage Process

128
ESN for Caching
 ESN model consists of
 Agents: Baseband units of a CRAN
 Input: the input is the users locations and context
information (e.g., requested videos, etc.)
 Output: the output is prediction of mobility patterns
 ESN model: This is the reservoir model, without going
through it now, it is composed of a set of matrices that enable
the recurrent neural network learning/predictions
 Conceptor: use of a week mobility as “pattern”
 For simulations, we use real data from BUPT and the
Youku video website

 Average transmit power of each UAV vs. number of users
 Using proposed approach, 20% reduction in transmit power
compared to other algorithms
Results
129

 The percentage users with satisfied QoE versus the number
of the users
 Using UAVs leads to a significant QoE improvement!
Results
130

 Decreasing transmit powers while increasing the number
of storage units
 UAV will directly transmit the requested contents to the users
Results
131

Other UAV Comm. Approaches
 UAVs as backhaul (see U. Challita and W. Saad, GLOBECOM
2017)
 UAV as cellular connected UEs (see works with U. Challita)
 More on machine learning (see M. Chen, W. Saad, et al.,
GLOBECOM 2017)
 UAVs as relay stations (see works by L .Swindelhurst et al. and
R. Zhang et al.)
 Deployment within a cloud radio access network and related
environments (see Yanikomeroglu et al.)
 Channel modeling, localization, tracking, public safety, and
related ideas (see works by I. Guvenc et al.)
132

134
CPS Security of UAVs
 UAVs are essentially cyber-physical systems
 Cyber vs. Physical: the physical world follows (typically)
laws of nature or control-theoretic models, which have
different constraints and time scales compared to cyber
features
 Human-in the loop: man meets machine (UAV)
 CPS nature brings cyber and physical vulnerabilities
 As UAVs become more prevalent, they will face more
and more security challenges
 Autonomy is both a blessing and a curse
 Let’s see an example security problem

Delivery Drones
 Drones will be used in the real-world for delivering goods
or to deliver rescue mission items
135

Security of Delivery Drones
 Delivery drones are prone to a variety of cyber-physical
security threats
 Cyber attacks to hack the cyber/wireless system and re-
route the drone or to jam its communication
 Commercial drones will be in the range of civilian-owned
hunting rifles that can be used for physical attacks
 In such scenarios, humans will be in the loop!
 Attackers will likely be humans (e.g., choose a high point
to shoot the drone or jam its link in line-of-sight)
 Vendors who own the drones will have stringent delivery
times especially for medical delivery (framing effects!) 136

Basic System Model
 A vendor sends a delivery drone from an origin O to a
destination D
 In an ideal world, vendor always chooses shortest path
 Presence of adversary
 Attackers can interdict the drone at several threat points
such as high buildings or hills to cause physical or cyber
damage
 A destroyed drone must be re-sent by the vendor, leading
to increased delivery times and economic losses
 The system can be modeled as a graph
137

Basic System Model
 The vendor is an evader wants to minimize expected
delivery time by choosing an optimal path
 The attacker is an interdictor who chooses a location to
attack the drone and maximize the delivery time
 Natural zero-sum network interdiction game 138

Game Formulation
 Two-player zero-sum game in which both vendor and
attacker want to randomize over their strategies
 Defender mixed-strategy vector
 Attacker mixed-strategy vector
 Attack at location n will be successful with probability pn
 The expected delivery time will be:
 fh(.) is a distance function
 T depends on various parameters 139

Game Formulation
 Vendor problem
 Adversary problem
 As a zero-sum game, it can be transformed into two
linear programs that can be easily solved
 Game admits a Nash (saddle-point) equilibrium
 There may be more than one equilibrium, but they are all
interchangeable yielding the same delivery time
 But what about the human perceptions? 140

141
Expected Utility Theory
 Conventionally, the Nash equilibrium is found
under expected utility theory (EUT)
considerations
 Presumes that players act rationally
 The players optimize the expected value over their mixed
strategies, i.e.,
 Caveat: in practice, it has been empirically shown that
when users are faced with uncertainty, they act
irrationally

142
Prospect Theory
 Lottery example
 Risk impacts
how players weigh
certain outcome
 Uncertainty can
lead players to deviate from the rational norms of EUT
 In CPS and UAV, many human players are in the loop and
will have subjective perceptions
 Solution: Prospect theory!
 W. Saad, A. Glass, N. Mandayam, and H. V. Poor, "Towards a Consumer-Centric Grid:
A Behavioral Perspective," Proceedings of the IEEE, vol. 104, no. 4, pp. 865-882, April
2016.

143
Prospect Theory
 Prospect theory
 Introduced by Kahneman and Tversky (1979)
 Won them the Nobel prize in 2002
 Cognitive psychology basis for analyzing human errors
and deviations from rational behavior
 Two important observations:
 Weighting effect: Players can subjectively weight
outcomes that are uncertain or risky
 Framing effect: Players may evaluate their utilities as
gains/losses with respect to a reference point

144
Illustrating the Weighting Effect
 Weighting effect
 Prelec function
 Outcomes are
weighted
differently
 Weighting applies
to
probabilistic
outcomes (e.g.
mixed strategies)

145
Prospect Theory
 With weighting, the players now optimize:
 Framing effect
 Each player will “frame” its gains/losses with
respect to a reference point
 Losses loom larger than gains
Weighting effect, Prelec function:

146
Prospect Theory
 Concave in
gains
 Convex in
losses
 Steeper slope
for losses as
opposed to gains
 Risk averse in
gains, risk
seeking in losses

147
Prospect Theory
 Framing effects
 The following framing function has been proposed:
 Suitable applications for PT?
 When humans are making decisions (CPS with human-in-
the loop, smart grid, pricing , human hackers, security)
 UAV security is a prime example, given the impact of
UAV performance on owners/humans

Prospect Theory in UAV
 The standard formulation does not account for the
presence of humans in the loop that are facing
uncertainty
 Uncertainty: perceptions of both attacker and vendor on
the probability of successful attack (weighting effect)
 Framing: subjective perception on the delivery time
with respect to a reference point
 Even the smallest of delays can be catastrophic
 For rescue situations, survival is at stake
 For Amazon, reputation can be damaged
148

Prospect Theory
 Subjective, PT-based utility
 The game is no longer zero-sum
 We consider max-min/min-max strategies
 Ongoing work to characterize equilibria under PT
149
Reference pointFraming function Weighting

150
Simulation results (1)
Due to the weighting
effect the vendor
will still choose
the shortest
path despite being
very risky (pn = 0.8)
This choice
becomes more likely
as the vendor
becomes more
irrational

151
Due to the weighting
effect, the attacker
focuses more
on nodes 5 and 8
which are
part of the shortest
path

152
Delivery time is increased by almost 10%
not accounting for time to re-load and
re-ship

153
As the loss
parameter increases
the vendor exaggerates
losses and thus
starts choosing more
risky paths to
meet delivery time
which, in turn,
yields to a
reverse effect!!!

Questions for the Future
 Will UAVs be the basis of future (6G?) cellular
networks, or will they remain relegated to temporary
coverage (hotspots, disaster)?
 Will UAV-UEs truly lead to fundamental new
challenges, or will it simply be a matter of
accommodating a new mobile node with a different
channel?
154

Conclusions
 UAVs provide with many new opportunities to
improve wireless communications
 The Internet of Flying Things will be upcoming and
we must be “analytically” ready
 Fundamental results on performance are needed
 Self-organization in terms of resources, network
topology, access modes, security, etc.
 Machine learning, game theory and related techniques
 Human-in-the-loop: bounded rationality
 Ubiquitous wireless connectivity from the sky!
155

Acknowledgement
 Acknowledgement to students and collaborators:
Mohammad Mozaffari, Anibal Sanjab, Mehdi Bennis,
Mingzhe Chen, Ursula Challita, Ismail Guvenc,
Mérouane Debbah, and others
156

157
Finally….
Thank You
Questions?

References
 M. Mozaffari, W. Saad, M. Bennis, and M. Debbah, “Unmanned aerial vehicle with
underlaid device-to-device communications: performance and tradeoffs,” IEEE
Transactions on Wireless Communications, vol. 15, no. 6, pp. 3949-3963, June 2016.
available online : https://arxiv.org/pdf/1509.01187.pdf
 M. Mozaffari, W. Saad, M. Bennis, and M. Debbah, “Efficient deployment of
multiple unmanned aerial vehicles for optimal wireless coverage,” IEEE
Communications Letters, vol. 20, no. 8, pp. 1647-1650, Aug. 2016.
 M. Mozaffari, W. Saad, M. Bennis, and M. Debbah, “Optimal Transport Theory for
Cell Association in UAV-Enabled Cellular Networks,” IEEE Communication Letters,
to appear, June 2017. available online : https://arxiv.org/pdf/1705.09748.pdf
 M. Mozaffari, W. Saad, M. Bennis, and M. Debbah, “Wireless Communication using
Unmanned Aerial Vehicles (UAVs): Optimal Transport Theory for Hover Time
Optimization,” Submitted to IEEE Transactions on Wireless Communications, to
appear, 2017, available online : https://arxiv.org/pdf/1704.04813.pdf
 A. Hourani, S. Kandeepan, and A. Jamalipour, “Modeling air-to-ground path loss for
low altitude platforms in urban environments,” in Proc. of IEEE Global
Telecommunications Conference, Austin, Tx, USA, Dec. 2014
158

References
 M. Mozaffari, W. Saad, M. Bennis, and M. Debbah, “Mobile Unmanned Aerial
Vehicles (UAVs) for Energy-Efficient Internet of Things Communications,” IEEE
Transactions on Wireless Communications, to appear, 2017, available online :
https://arxiv.org/pdf/1703.05401.pdf
 M. Mozaffari, W. Saad, M. Bennis, and M. Debbah, “Mobile Internet of Things: Can
UAVs provide an energy-effcient mobile architecture?,” in Proc. of IEEE Global
Communications Conference (GLOBECOM), Washington, DC, USA, Dec. 2016.
 M. Mozaffari, W. Saad, M. Bennis, and M. Debbah, “Drone small cells in the clouds:
Design, deployment and performance analysis,” in Proc. of IEEE Global
Communications Conference (GLOBECOM), San Diego, CA, USA, Dec. 2015.
 M. Mozaffari, W. Saad, M. Bennis, and M. Debbah, “Optimal transport theory for
power efficient deployment of unmanned aerial vehicles," in Proc. of IEEE
International Conference on Communications (ICC), Kuala Lumpur, May 2016.
159

References
 M. Chen, M. Mozaffari, W. Saad, C. Yin, M. Debbah, and C. S. Hong, “Caching in the
Sky: Proactive Deployment of Cache-Enabled Unmanned Aerial Vehicles for Optimized
Quality-of-Experience,” IEEE Journal on Selected Areas in Communications (JSAC),
Special Issue on Human-In-The-Loop Mobile Networks, vol. 35, no. 5, pp. 1046-1061,
May 2017. available online: https://arxiv.org/pdf/1610.01585.pdf
 M. Naderisoorki, M. Mozaffari, H. Manshaei, and H. Saidi, “Resource Allocation for
Machine-to-Machine Communications with Unmanned Aerial Vehicles," in Proc. of
IEEE GLOBECOM Workshop, Washington, DC, USA, Dec. 2016.
 D. Athukoralage, I. Guvenc, W. Saad, and M. Bennis, "Regret Based Learning for UAV
assisted LTE-U/WiFi Public Safety Networks," in Proc. of the IEEE Global
Communications Conference (GLOBECOM), Mobile and Wireless Networks Symposium,
Washington, DC, USA, Dec. 2016.
available online : http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=7842208
 A. Sanjab, W. Saad, and T. Başar, "Prospect Theory for Enhanced Cyber-Physical
Security of Drone Delivery Systems: A Network Interdiction Game," in Proc. of the
IEEE International Conference on Communications (ICC), Communication and
Information Systems Security Symposium, Paris, France, May 2017.
160

References
 W. Saad, Z. Han, T. Başar, M. Debbah, and A. Hjorungnes, “Hedonic Coalition
Formation for Distributed Task Allocation among Wireless Agents,” IEEE Transactions
on Mobile Computing, vol. 10, no.9, pp. 1327-1344, Sep. 2011.
 U. Challita and W. Saad, "Network Formation in the Sky: Unmanned Aerial Vehicles for
Multi-hop Wireless Backhauling", in Proc. of the IEEE Global Communications
Conference (GLOBECOM), Singapore, December 2017.
 M. Chen, W. Saad, and C. Yin, "Liquid State Machine Learning for Resource Allocation
in a Network of Cache-Enabled LTE-U UAVs", in Proc. of the IEEE Global
Communications Conference (GLOBECOM), Singapore, December 2017.
 M. Chen, U. Challita, W. Saad, C. Yin, and M. Debbah, "Machine Learning for Wireless
Networks with Artificial Intelligence: A Tutorial on Neural Networks",
arXiv:1710.02913, 2017, online: https://arxiv.org/pdf/1710.02913.pdf
 Y. Zeng, R. Zhang, and T. J. Lim, “Throughput maximization for UAV-enabled mobile
relaying systems,” IEEE Transaction on Communications, vol. 64, no. 12, pp. 4983–
4996, Dec. 2016.
161

Tutorial on Wireless Communications and Networking with Drones and Unmanned Aerial Vehicles (UAVs)

More Related Content

What's hot (20)

Similar to Tutorial on Wireless Communications and Networking with Drones and Unmanned Aerial Vehicles (UAVs) (20)

Recently uploaded (20)

Tutorial on Wireless Communications and Networking with Drones and Unmanned Aerial Vehicles (UAVs)