IEEE - Aerospace and Electronic Systems - October 2022 - 34

Networks of UAVs of Low Complexity for Time-Critical Localization
Q-table, need to be delegated to the edge. In fact, given the
high-dimensional representation of the environment and the
presence of multiple UAVs with different views of it, the
edge served as a leader for fusing the received information,
storing and updating a globalQ-table. Thanks to this centralized
approach, both agents had better ambient awareness,
and they could infer a global policy according to Q-learning.
PERFORMANCEVERSUSTIME-CRITICAL
In the next, we assess the detection performance as a function
of time and number of UAVs. We omit a discussion
about the convergence to an optimal policy, which is of
secondary importance for time-critical applications as it
would require lengthy training procedures [64].
Figure 7 shows the UAV trajectories with blue and
yellow markers for two UAVs, and for the first and last
episodes. As expected, the UAV named " A, " whose position
was closer to the target, got sooner to the destination
but with a scarce map reconstruction. By contrast, the
UAV " B " was capable of reconstructing a better copy of
the map, and of finding a good trajectory after some training
episodes thanks to the aid offered by the other UAV.
In the next, we assess the learning performance in
terms of detection rate, defined as
Zk ¼
1
NMC
NXMC
m¼1
I9SNRk >h
(2)
where SNR is the vector of the SNRs experienced by the
UAVs, h is a threshold set to 15 dB to guarantee a reliable
detection rate (i.e., 80%), and k is the time accounting
for both episodes and time instants. The logical condition,
i.e., 9SNRk > h, is satisfied if at least one UAV receives
a power such that the SNR overcomes the threshold that
allows to correctly detect the target.
Figure 8 shows the percentage of times the target is
detected [e.g., Zk in (2)] through time and for different numbers
ofUAVs. As an example, for N ¼ 4, the target is considered
detected in the 60%-70% of times within the first
episode (e.g., Time ¼ 1)whereas,for N ¼ 1,such a value
is much lower. Notably, when multiple UAVs update a common
Q-table, the required time in terms of episodes is less
than the single agent case. This result confirms that the
employment of many coordinated UAVs allows to accelerate
the learning procedure, thus enabling the exploration of
large environments, which is crucial for time-critical applications.
On the other side, when the network is composed of
many UAVs, the consequent communication burden with
the edge should be accounted for.
CONCLUSIONS AND FUTURE PERSPECTIVES
It is clear that UAVs present a wide range of requirements:
they need to be autonomous, fast, multi-tasking, and
34
Figure 7.
Reference map is in subfigure (a): Red and green markers indicate
the initial positions ofUAVs and the target position, respectively.
Examples of estimated trajectories and maps for the first (b), (d)
and second agent (c), (e) and for the first (b), (c) and last (d), (e)
episodes.
intelligent. But, above all, they should preserve a low-complexity
implementation while accurately accomplishing the
mission for which they have been deployed. Consequently,
the onboard UAV-I needs to be properly designed and, when
necessary, delegated to a higher entity of the network thanks
to a communication scheme enabling such a distribution. In
this sense, even if the investigated case studies depend on the
specific setup, some general conclusions can be drawn.
FINALREMARKSONCOMMUNICATIONS
In Use Case 1, we considered both edge-based and peer-topeer
multi-hops solutions. In the edge scenario, the availability
of a global view of the environment naturally improves
the accuracy of the estimation, but the performance depends
on the latency due to access and processing delays. Indeed,
the presence ofaged measurements might lead to an old view
ofthe system state, especially in fast-changing environments.
In the case of multi-hops, the delays are dictated by the
connections created at each time instant between the nodes.
Therefore, if the network is small and can be connected with
IEEE A&E SYSTEMS MAGAZINE
OCTOBER 2022
IEEE - Aerospace and Electronic Systems - October 2022

Table of Contents for the Digital Edition of IEEE - Aerospace and Electronic Systems - October 2022
