Command Control Uav
Command Control Uav
Command Control Uav
net/publication/323609794
CITATIONS READS
2 1,826
3 authors, including:
Some of the authors of this publication are also working on these related projects:
Optimized algorithms for communications trough Fly ad-Hoc Network View project
All content following this page was uploaded by Pietro Cassara on 04 February 2019.
1.1 Introduction
Nowadays, UAVs, or drones, are attracting a lot of attention from industrial and
research fields. They are suited to a large number of applications, from military
to civilian ones, thus the advantages that they can provide are eagerly exploited
by several actors. UAVs have been largely used in the military field in the past,
but, nowadays, the cost reduction makes them of interest also in several civilian
fields: for instance, in the precision agriculture [1, 2], in the surveillance field or
environmental monitoring [3], and in search and rescue applications [4]. The use
of UAVs has proved to be particularly effective in otherwise impervious areas,
or each time their use can remove the need for expensive temporary scaffolding,
such as in the case of the inspection of historical or cultural areas and buildings.
In the latter scenarios, UAVs are typically equipped with the needed sensors
in order to facilitate an inspection: for instance, cameras, but also short-range
communication radios, in order to collect data from previously installed sensors
or to deliver commands, in the case of actuators. Up to now, the largest fraction
of civilian applications is based on the use of Line of Sight (LoS) communications,
so that the operator remotely controlling the drone can avoid any close obstacles.
In fact, a strict regulation is quickly spreading in several European countries, in
order to control the use of these devices, mainly in areas where poor experienced
personnel can improperly use UAVs, such as close to airports or in the presence
of crowd, with possibly disastrous consequences. When considering Non Line of
Sight (NLoS) or Beyond Line of Sight (BLoS) communications, the use of the
satellites is a possibility, but the following limiting factors should be taken into
account: a larger delay than in LoS communications; the absence of a direct visual
2 Pietro Cassará, Marco Colucci, Alberto Gotta,
feedback; the need of always available bandwidth, in order to control the drone
and to collect data; furthermore, the availability of automatic collision avoidance
systems to compensate the operators’ maneuvering delay. The aforementioned
requirements make more expensive the design, the manufacturing, and the use of
these devices. In several contexts, the use of a single UAV can be a limiting factor;
for instance, in search and rescue applications, if several drones can be rapidly
deployed, the probability of a successfully rescue mission may increase. The use
of UAV swarms is of interest in several fields, if the task previously assigned
to a single drone can be parceled and parallelized. A number of advantages
are provided by the use of UAV swarms, as pointed out in [5]: (i ) likely, the
overall cost of acquisition and maintenance of several small Commercial Off-the-
Shelf (COTS) UAVs is lower than the overall cost of a single large UAV; (ii )
scalability, which is a key feature of UAV swarms, instead absent in single UAV
missions; (iii ) fault-tolerance, because a single malfunction has a limited impact
on the swarm; (iv ) faster operations, thanks to the parallelization of the work.
The scenario under consideration in this work is built upon UAV swarms
remotely controlled via satellite. Although some works in the literature deal with
the use of the drone swarms [6, 7, 8], the issues posed by Command and Control
(C2) via satellite require further investigations. The main contribution of this
work is providing an analytical framework to estimate both coverage probability
and delivery delay, when an UAV swarm receives C2 data via satellite, which
can be further forwarded (gossiped ) inside the UAV swarm, in order to increase
the probability of a reliable data delivery. The rest of the paper is structured as
follows: Section 1.2 provides some background and discusses the related works.
Section 1.3 deals with the description of the problem and of the analytical model
needed to address it. Section 1.4 provides some preliminary numerical results;
the conclusions are in Section 1.5.
In the scenario under consideration, visible in Figure 1.1a, a Ground Control Sta-
tion (GCS) transmits C2 messages, each composed of k control blocks, towards
a swarm composed of n drones. Therefore, we assume that each C2 message (for
instance, new navigation data) is split in k fragments that must be successfully
received. An average loss probability P LSAT is considered, in order to take into
Fig. 1.1
1
DeSIRE stands for Demonstration of Satellites enabling the Insertion of RPAS in
Europe, a joint ESA-EDA initiative.
4 Pietro Cassará, Marco Colucci, Alberto Gotta,
where Ptr = 1−(1−τ )n is the probability that there is at least one transmission in
a time-slot; Ps = (n τ (1−τ )n−1 )/(1−(1−τ )n ) is the probability of a successfully
transmission on the channel, given that at least one node has transmitted; Tid is
the duration of a time-slot; L is the payload size; Ts = M ACheader +L+SIF S +
2Tid + ACK + DIF S is the average duration of the busy period of the channel
because of a successful transmission; Tc = M ACheader + L + DIF S + Tid is the
average duration of the busy period of the channel because of a collision. SIF S,
DIF S and ACK parameters in use in this work are provided in Table 1.1.
By using (1.1), we can compute the delivery delay of the gossiped blocks
on the 802.11 channel. We are interested in evaluating the coverage probability
1 Command and Control of UAV Swarms via Satellite 5
P cov , which is the probability that each node receives all the k blocks composing
a C2 message. It is defined by the following equation:
Equation (1.2) is composed of two terms: the probability PSAT that a node
successfully receives the blocks via satellite, and the probability PFcovAN that the
blocks are successfully received from the gossiping neighbors. In order to estimate
PFcov
AN , we need to enumerate the coverage events. A coverage event occurs if all
k blocks are received by a single UAV thanks to gossiping. In order for a coverage
event to occur, the UAV must receive at least once each of the k control blocks2 .
Assuming that a coverage event has occurred, the total number of control blocks
received by the UAV, namely h, is bounded as follows: k ≤ h ≤ k (n − 1).
In order to enumerate the coverage events, we need the following definitions: a
row (column) margin is defined as the sum of the entries, rows by rows (columns
by columns). The coverage events can be enumerated by counting the number
of receiving matrices M(R(h), C(h)). A receiving matrix is composed of vectors
R(h) = {r1 (h), , ..., rn (h)} and C(h) = {c1 (h), , ..., ck (h)}, which are the row
and column matrix margins, respectively, for a given h. Matrices M(R(h), C(h))
are (n − 1) × k binary matrices, where the entry (i, j) is 1 or 0, if the i-th
neighbor successfully transmits (or not) the j-th block. An example of a receiving
matrix M is shown in Figure 1.1b(a), for k = 4 and n = 5, with margins
R(h) = {4, 2, 0, 0, 0}, C(h) = {2, 2, 1, 1}. The sum of the elements of R(h) and
C(h) in Figure 1.1b(a) is h = 6. The set of margins C(h), corresponding to the
coverage events, can be enumerated by evaluating all the k integer partitions
of h with h = k, · · · , k (n − 1). We recall that the integer partitions of h are
the ways of writing h as a sum of k positive integers C(h) = {c1 (h), · · · , ck (h)}.
Some of thePpartitions may not be feasible; a partition is said to be feasible if
and only if i ci (h) ≥ k and ci (h) ≥ 1, for i ∈ [1, k]. In order to enumerate only
the feasible partitions, some constraints are needed. Those constraints can be
written as follows:
k
X
h= ci (h), ∀ C(h) ∈ C(h), h = k, · · · , k(n − 1); (1.3)
i=1
1 ≤ ci (h) ≤ n − 1, ∀ C(h) ∈ C(h).
So far, we are able to enumerate all the possible coverage events for the parame-
ters h and the number of gossiping users Utx . Finally, the occurrence probability
of a coverage events is:
In other words, (1.5) allows evaluating the probability that an UAV receives
h blocks given Utx transmitting neighbors. The remaining (n − 1 − Utx ) UAVs do
not contribute to h, because the blocks transmitted via satellite and via gossiping
have been lost. Let P r(Utx ) be the probability of having Utx gossiping neighbors.
The coverage probability can be written as:
X
PFcov
AN = P r(h|Utx ) P r(Utx ),
h
h in−Utx −1
Utx
P r(Utx ) = (PSAT ) (1 − PSAT ) + PSAT P LkF AN .
Figure 1.3 shows that the coverage probability vs. the number of gossiping
neighbors, when a C2 packet composed of k = 4 blocks is sent. The use of a
gossiping algorithm can significantly increase the coverage probability. However,
when considering severe impairments on the satellite channel (P LSAT = 0.1),
the coverage cannot be guaranteed, as shown in 1.3 with 10 UAVs. Different
approaches, such as either forward error correction techniques for real-time traffic
8 Pietro Cassará, Marco Colucci, Alberto Gotta,
as in [12, 13], or the use of Network Coding (NC) as in [14], can be employed to
increase the coverage probability; such investigation is left out for future works.
0.045
0.04
0.25
GEO PLsat =0.01
0.035
GEO PLsat =0.05
gossiping delivery delay [s]
0.01
PLsat =0.1 0.05
0.005 PLsat =0.05
PLsat =0.01
0 0
2 4 6 8 10 2 4 6 8 10
number of nodes number of nodes
(a) 802.11 delivery message delay vs. (b) Overall message delivery delay vs.
number of gossiping nodes number of gossiping nodes
Fig. 1.2
0.95
0.9
coverage probability
0.85
0.8
0.75
0.7
PLSAT=0.1
0.65 PLSAT=0.05
PLSAT=0.01
0.6
2 4 6 8 10
number of nodes
Figure 1.2a shows the impact (defined as gossiping delay) of increasing the
number of gossiping nodes on the average delivery delay, due to 802.11 backoff
mechanism, which reduces the time a node contends for the medium, to reduce
the collision probability. Figure 1.2b shows the one-way-delay for delivering a
C2 message from the GCS to the UAV swarm, in case of GEO, MEO and LEO
satellite. This delay is the sum of two components: the satellite latency, weighted
by PSAT , and the gossiping delay (in Figure 1.2a), when the message is received
via gossiping. The use of a gossiping algorithm does not significantly impact on
1 Command and Control of UAV Swarms via Satellite 9
the delivery delay; in fact, the largest component is due to the latency of the
satellite. However, the channel latency is still higher than the gossiping delay in
case of LEOs, but the latter is no more so negligible w.r.t. the former.
1.5 Conclusions
Acknowledgments
This work has been supported by the Tuscany region in the frameworks of SCI-
ADRO and MOSCARDO projects (FAR-FAS 2014-2016).
References
1. M. Bacco, E. Ferro, and A. Gotta, “Radio propagation models for uavs: what is
missing?” in Proceedings of the 11th International Conference on Mobile and Ubiq-
uitous Systems: Computing, Networking and Services. ICST (Institute for Com-
puter Sciences, Social-Informatics and Telecommunications Engineering), 2014, pp.
391–392.
2. F. M. Bacco, E. Ferro, and A. Gotta, “UAVs in WSNs for agricultural applications:
an analysis of the two-ray radio propagation model,” in IEEE SENSORS 2014
Proceedings. IEEE, 2014, pp. 130–133.
3. G. Chmaj and H. Selvaraj, “Distributed processing applications for UAV/drones:
a survey,” in Progress in Systems Engineering. Springer, 2015.
4. S. D’Auria, M. Luglio, C. Roseti, R. Strollo, and F. Zampognaro, “Real Time
Transmission of Cultural Heritage 3D Survey in Case of Emergency,” in 3rd Inter-
national Conference on Information and Communication Technologies for Disaster
Management, Vienna,Austria, December 2016.
5. I. Bekmezci, O. K. Sahingoz, and Ş. Temel, “Flying ad-hoc networks (FANETs):
A survey,” Ad Hoc Networks, vol. 11, no. 3, pp. 1254–1270, 2013.
6. E. W. Frew and T. X. Brown, “Networking issues for small unmanned aircraft
systems,” Journal of Intelligent and Robotic Systems, vol. 54, no. 1-3, 2009.
7. J. J. Corner and G. B. Lamont, “Parallel simulation of UAV swarm scenarios,”
in Proceedings of the 36th conference on Winter simulation. Winter Simulation
Conference, 2004, pp. 355–363.
10 Pietro Cassará, Marco Colucci, Alberto Gotta,