=Paper=
{{Paper
|id=Vol-1867/w12
|storemode=property
|title=An Optimization-Based Algorithm for Indoor Localization of JADE Agents
|pdfUrl=https://ceur-ws.org/Vol-1867/w12.pdf
|volume=Vol-1867
|authors=Stefania Monica,Federico Bergenti
|dblpUrl=https://dblp.org/rec/conf/woa/MonicaB17a
}}
==An Optimization-Based Algorithm for Indoor Localization of JADE Agents==
https://ceur-ws.org/Vol-1867/w12.pdf
65
An Optimization-Based Algorithm
for Indoor Localization of JADE Agents
Stefania Monica, Federico Bergenti
Dipartimento di Scienze Matematiche, Fisiche e Informatiche
Università degli Studi di Parma
Parco Area delle Scienze 53/A, 43124 Parma, Italy
Email: {stefania.monica,federico.bergenti}@unipr.it
Abstract—This paper describes and evaluates an optimization- a bidirectional link with the physical environment, but they
based localization algorithm which has been recently imple- do not normally provide means to let agents know their
mented to enrich the possibilities of the localization add-on mod- position in indoor environments. While localization can be
ule for JADE. The described algorithm targets indoor scenarios
and it enables localization of JADE agents running on smart considered a solved problem in outdoor environments because
devices in known environments, provided that a conventional consolidated technologies like the Global Positioning System
WiFi network is present. The algorithm assumes that WiFi access (GPS) are commonly available in virtually all smart devices,
points in fixed and known positions are available, and it estimates only few smart devices offer localization sensors for indoor
the position of the smart device where the agent is running using environments. Just to mention one of the most promising
estimates of the distance between the smart device and each
access point. Distance estimates are used to build an optimization possibilities in this respect, which is already available in
problem, whose solution is an estimate of the position of the smart some smart devices, the Ultra-Wide Band (UWB) technology
device. The described algorithm uses particle swarm optimization guarantees accurate and robust indoor localization [4], and
to solve the built optimization problem, but it is open to the it has been already used for accurate indoor localization in
adoption of other optimization techniques. The validity of the industrial environments [5]–[7] and to provide agents with
proposed approach is supported by experimental results shown
in the last part of the paper. indoor localization capabilities [8]. However, besides the need
of a specific type of smart device, the strongest limitation
I. I NTRODUCTION that we currently see in the adoption of UWB or similar
Among the longstanding debate on the characteristics that technologies is that they require a dedicated infrastructure,
should be ascribed to agents, the fact that agents should be which needs to be implemented in indoor environments just
considered situated entities immersed in a partially observable to support localization. Notably, WiFi networks can be con-
environment seems consolidated. Agents acquire knowledge sidered ubiquitous in indoor environments today, and the use
from the environment, and they act on it to achieve their of standard WiFi infrastructures to support localization, rather
goals. Whether the environment is physical or not is irrele- than dedicated infrastructures, can be considered a major
vant for a generic characterization, but agents executing in advancement and an enabler for applications [9], [10].
physical environments are traditionally called robots, while In this paper we document recent developments of experi-
agents executing in nonphysical environments are traditionally ments intended to provide JADE agents running on common
called software agents. Such a distinction is quickly fading Android devices with localization capabilities in known indoor
because onboard software of robots has now features that environments using only ordinary WiFi networks. In detail, the
were typical of software agents (see, e.g., [1]), and smart targeted scenario assumes that a JADE agent is interested in
devices [2] provide software agents with a direct link with accurate and timely estimates of its dynamic position within
the physical world that should be taken into serious account an indoor environment where known and static WiFi Access
to effectively implement the envisioned ideas of agent-based Points (APs) are available. The proposed method is based on
ubiquitous and pervasive computing. Today, smart devices the possibility of acquiring estimates of the distance between
are ideal candidates to host JADE containers with minimal the smart device where the agent is running and each AP
restrictions [3], and it is perfectly reasonable to assume that using the received power of signals from responding APs
JADE should provide platform-level functionality to let agents during standard network discovery. Such distance estimates are
access the bidirectional link with the physical world that smart then processed to estimate the position of the smart device in
devices offer. In order to fully comply with the metaphor the indoor environment. Once computed, position estimates
of agents as situated entities, JADE is demanded to provide are immediately made available to the JADE agent. The
agents with platform-level functionality to let agents sense the processing of distance estimates can be performed using one
physical environment where they execute and act on it to bring of the localization algorithms available in the literature [11],
about their goals. but in this paper we describe and evaluate the features of an
Smart devices already integrate sophisticated sensors and algorithm that was introduced in [12] and further improved
actuators that can be effectively used to provide agents with in [13], [14] to overcome the inherent numerical instability
66
of classic localization algorithms. The described algorithm, always provided by operating systems to applications during
which was originally designed to use UWB, turns the local- standard WiFi network discovery because it is normally used
ization problem into a specific optimization problem, which to let the user choose among available networks. Each range
is solved using Particle Swarm Optimization (PSO), even if estimate can also be associated with the corresponding AP
other optimization techniques could be adopted. and, eventually, with its coordinates, because communications
This paper is organized as follows. Section II fixes notation between the TN and an AP during network discovery include
and describes the implemented algorithm. Section III shows the Basic Service Set IDentification (BSSID) of the latter,
experimental results obtained in a representative indoor sce- which can be used to identify the responding AP. Hence,
narios. Finally, Section IV concludes the paper and outlines assuming that each known BSSID can be associated with
possible future developments. the coordinates si of the corresponding AP, each distance
estimate can be related to the coordinates of the corresponding
II. O PTIMIZATION - BASED L OCALIZATION
AP. Notably, needed information to support the computation
The described algorithm is implemented in a localization of distance estimates is always available to applications and
add-on module for JADE, which has been recently imple- neither low-level access to hardware, nor privileged operating
mented to provide a framework to host localization algorithms, system services are needed. Indeed, the TN is not even
and to offer agent developers the possibility of choosing requested to be connected to the WiFi network, because only
among different algorithms to address the specific needs of network discovery is used.
applications [10]. A discussion of the architecture of this We have already discussed the performance of some local-
module, which is briefly summarized in [8], is not needed ization algorithms in the context of agent-based indoor local-
to describe the algorithm and its features. ization, such as the circumference intersection algorithm [10]
A. Notation and Reference Scenarios and the two-stage maximum-likelihood algorithm [8]. In this
paper, we focus on a localization approach based on optimiza-
The smart device where the agent is running, which is
tion, where the localization problem is rewritten in terms of an
denoted as Target Node (TN) in the rest of the paper, is situated
optimization problem. In order to describe the algorithm, let us
in an indoor environment that contains M WiFi APs. Let
first introduce proper notation and makes relevant geometric
si = (xi , yi )T i ∈ {1, . . . , M }. (1) considerations. Let
u = (x, y)T (3)
be the coordinates of such APs. We assume that APs are static
and we also assume that their coordinates si are known to be the true position of the TN, which is supposed to be
the agent. Note that, in order to simplify notation, we focus unknown and which is what should be estimated. The true
on the description of the algorithm on bi-dimensional scenar- distance between the TN and the i−th AP can be denoted as
ios. The generalization of the described algorithm to three-
ri , ||u − si || i ∈ {1, . . . , M }. (4)
dimensional scenarios is straightforward. Moreover, under the
assumption that the smart device approximatively moves on a The knowledge of true distances {ri }M i=1 , together with the
plane, [10] shows how to relate a localization problem in a knowledge of coordinates {si }M i=1 of the APs, would easily
three-dimensional scenario to a proper localization problem in determine the position of the TN. In this case, the coordinates
a bi-dimensional scenario. of the TN could be found by simply intersecting the circum-
The ranging capabilities that the smart device is requested ferences centered in {si }M i=1 , whose radii are {ri }i=1 . This
M
to offer in order to support the implemented algorithm concern translates into the following system of M quadratic equations
the possibility of measuring estimates of the distances to 2 2 2
the M APs. Such estimates of distances are obtained by (x − x1 ) + (y − y1 ) = r1
processing the average received power of the WiFi signals ... (5)
traveling between the TN and each responding AP during (x − xM )2 + (y − yM )2 = rM 2
.
standard network discovery. According to the Friis transmis-
sion equation [4], the average received power P̄ (r) can be Unfortunately, since true distances {ri }Mi=1 between APs
expressed as a function of the distance r between a transmitter and the TN are unknown, localization can only be performed
and a receiver. By inverting the Friis transmission equation, the using the following system of quadratic equations
value of r as a function of P̄ (r) can be expressed as 2 2 2
(x̂ − x1 ) + (ŷ − y1 ) = r̂1
P̄ (r)−P0
r = r0 · 10− 10β (2) ... (6)
(x̂ − xM )2 + (ŷ − yM )2 = r̂M
2
where P0 is the known power at reference distance r0 , and
β depends on the characteristics of the transmission. Hence, which is obtained from (5) by replacing the values of true
in order to derive an estimate of the distance r between the distances {ri }M
i=1 , with their estimates, denoted as {r̂i }i=1 .
M
TN and a generic AP, it is sufficient to measure the average Due to errors on distance estimates, the M circumferences
received power of the signal traveling between them and to corresponding to the equations in (6) often do not intersect
apply (2). Such a measure of the average received power is in a single point and, for this reason, a proper localization
67
algorithm needs to be considered in order to find the proper move all the particles towards the position corresponding to
estimates of the position of the TN, which is denoted as the optimal solution of the considered minimization problem.
The PSO-based algorithm that we adopted to solve the min-
û = (x̂, ŷ)T . (7) imization problem (10) works as follows. First, the positions
In order to derive a proper localization algorithm, let us first of the particles are randomly initialized in the search space,
observe that (6) can be re-written in matrix notation as which, in our context, corresponds to the physical indoor
environment where the APs and the TN are situated. The initial
1 ûT û + A û = k̂ (8) positions are denoted as
where 1 is vector with M elements equal to 1, k̂ is a vector x(i) (0) i ∈ {1, . . . , S} (12)
whose i−th element is r̂i2 − (x2i + yi2 ), and A is the following where i is the index of the generic particle, and S is the number
M × 2 matrix of particles. Analogously, the velocity of the i−th particle is
x1 y1 initialized with the value
x2 y2
v (i) (0) i ∈ {1, . . . , S}. (13)
A , −2 . .. . (9)
.. .
After the initialization phase, positions and velocities of all
xM yM
the particles are updated at each iteration t ∈ N to simulate
Possible algorithms to solve (8), based on least square interactions among individuals. More precisely, at the t−th
techniques, on Taylor series expansion, or on maximum- iteration, the velocity of the i−th particle whose position is
likelihood methods, can be found, for instance, in [15]. The x(i) (t) is updated according to the following rule [17]
rest of this section describes an alternative algorithm, which
v (i) (t + 1) = ω(t)v (i) (t)
was introduced to overcome numerical instability problems of
mentioned approaches. + c1 R1 (t)(y (i) (t) − x(i) (t)) (14)
+ c2 R2 (t)(y(t) − x(i) (t))
B. The PSO-Based Localization Algorithm
Various algorithms to solve localization problems expressed where, as discussed in [18],
as systems of equations like (8) can be found in the significant 1) y (i) (t) is the best position reached so far;
body of literature on the subject (see, e.g., [15] and referenced 2) y(t) is the best position globally reached so far;
literature). All such algorithms typically suffer from numerical 3) ω(t) is the so called inertial factor;
instability in correspondence of peculiar configurations of 4) c1 is a positive parameter called cognition parameter;
network nodes in space, for example if APs are aligned as 5) c2 is a positive parameter called social parameters; and
discussed in [14]. In order to derive a more robust algorithm, 6) R1 (t) and R2 (t) are independent random variables uni-
in [14] it is proposed to reformulate (8) as an optimization formly distributed in (0, 1).
problem, which is solved using Particle Swarm Optimization From (14) it can be easily observed that the velocity of a
(PSO) [16], as outlined in the rest of this section. Note that the particle at the (t + 1)−th iteration is obtained as the sum of
proposed PSO-based approach was originally designed to use three addends. The first addend is related to the velocity of the
UWB signaling, and the description of its evolution to WiFi particle at the previous iteration, which is weighed according
signaling is the major contribution of this paper. to the inertial factor ω(t). The second addend is meant to move
Observe that (8) can be written as a minimization problem each particle towards the best position it reached so far. Note
whose solution is an estimate of the position of the TN that such a best position is the one which corresponds to the
lowest value of the fitness function and, therefore, y (i) (t) can
û = arg min F (u) (10) be expressed as
u
where F (u) represents the fitness function associated with the y (i) (t) = arg min F (z) (15)
z∈X (i) (t)
problem, which is defined as
where
F (u) = ||k̂ − (1 ûT û + A û)||. (11) X (i) (t) = {x(i) (0), . . . , x(i) (t)}. (16)
In order to solve the minimization problem (10), among Finally, the third addend aims at moving each particle towards
the wide range of possibilities, we propose to use the PSO the global best position, which is the position that corresponds
algorithm for its proved effectiveness and robustness. Accord- to the smallest value of the fitness function among all those
ing to such an algorithm, the set of potential solutions of reached by any particle in the swarm [19]. Hence, y(t) is
a minimization problem can be considered as a swarm of expressed as
particles whose positions and velocities are iteratively updated y(t) = arg min F (z) (17)
according to proper rules. Such rules are inspired by biological z∈Y (t)
phenomena like the movements of birds in swarms. In the where
context of optimization problems, such rules are meant to Y (t) = {y (1) (t), . . . , y (S) (t)}. (18)
68
3 3
2.5 AP2 AP4 2.5
2 2
1.5 1.5
y[m]
y[m]
1 1
0.5 0.5
TN TN
0 0
−0.5 AP1 AP3 −0.5 AP1 AP2 AP3 AP4
−1 −1
−1 0 1 2 3 4 5 6 −1 0 1 2 3 4 5 6
x[m] x[m]
(a) (b)
Fig. 1. The positions of the four APs (blue squares) and of the TN (red Fig. 2. The positions of the four APs (blue squares) and of the TN (red
star) of the first scenario are shown in considered room. The walls of the star) of the second scenario are shown in considered room. The walls of
room are shown in black. the room are shown in black.
Typically, the inertial factor ω(t) is chosen as a decreasing a three-dimensional scenario to a proper localization problem
function of t to guarantee low dependence of the solution on in a bi-dimensional scenario.
the initial population, and to reduce the exploration ability of We assume that M = 4 APs are used to estimate the posi-
the swarm as the number of iterations increases, which makes tion of the TN and, in the same environment, we consider two
the method similar to a local search in last iterations [17]. different configurations of such APs. The two configurations
The velocities computed with (14) are used to update are shown in Fig. 1 and in Fig. 2, respectively. In this figures,
the positions of particles at each iteration according to the the positions of the APs are marked with blue squares, the
following rule position of the TN is marked with a red star, and the black
lines show the walls of the corridor. Note that, even if the
x(i) (t + 1) = x(i) (t) + v (i) (t) i ∈ {1, . . . , S}. (19) algorithm is evaluated for M = 4 APs, it can be executed with
From (19) it can be observed that the position of the i−th an arbitrary number of APs. The minimum number of APs to
particle at the (t+1)−th iteration is simply obtained by adding localize a TN in a bi-dimensional environment is three, but the
v (i) (t) to its previous position. availability of more APs contributes to improve the accuracy
The execution of the PSO algorithm terminates when a of localization.
proper termination condition is met. Normally, the termination For each configuration of APs, the optimization-based local-
condition includes that a sufficient number of iterations was ization algorithm outlined in Section II is applied 100 times,
performed, or that the fitness function reached a satisfactory thus leading to 100 independent position estimates for the TN.
value. Once the execution of the algorithm terminates, the In the following, the position estimates of the TN at the j−th
solution of the minimization problem is the position of the iteration are denoted as
particle with the lowest value of the fitness function.
û(j) = (x̂(j) , ŷ (j) ) j ∈ {1, . . . , 100}. (20)
The PSO algorithm outlined above is used to solve the
localization problem (10). In order to obtain the experimental The performance of the proposed localization algorithm is
results shown in the next section, we set the value of the analyzed in terms of a localization error computed as the
inertial factor to 0.5. The values of c1 and c2 are equal and distance between each position estimate û(j) and the true
they are set to 2, so that the average values of c1 R1 (t) and position of the TN, denoted as u, which is known. Such a
of c2 R2 (t) correspond to 1. The size of the population S is localization error is
set to 40, and the algorithm terminates after 50 iterations.
These values proved to be effective for localization purposes d(j) , ||û(j) − u|| j ∈ {1, . . . , 100}. (21)
as documented in the experimental results presented in [12].
The values of {d(j) }100
j=1 can be used to compute the average
III. E XPERIMENTAL R ESULTS localization error, denoted as davg , which is expressed as
This section shows experimental results obtained using the 1 X (j)
100
described optimization-based algorithm, as implemented in davg = d . (22)
100 j=1
the localization add-on module for JADE. Presented results
are obtained in a representative scenario which consists in a We start by considering experimental results obtained in the
section of a corridor whose width is 2 m. Note that we consider scenario shown in Fig. 1. In this case, the coordinates of the
a bi-dimensional scenario, however, the algorithm can be APs, expressed in meters, are
generalized to three-dimensional scenarios as shown in [20].
Moreover, under the assumption that the smart device moves s1 = (0, 0)T s2 = (0, 2)T
(23)
on a plane, [10] shows how to relate a localization problem in s3 = (5, 0)T s4 = (5, 2)T .
69
2 2
davg davg
1.8 1.8
1.6 1.6
1.4 1.4
1.2 1.2
d ( j ) [m]
d ( j ) [m]
1 1
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0 0
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
j j
Fig. 3. The values of the localization error d(j) is shown for 100 position Fig. 4. The values of the localization error d(j) is shown for 100 position
estimates for the configuration of APs in Fig. 1. The value of the average estimates for the configuration of APs in Fig. 2. The value of the average
localization error is shown with a magenta line across the diagram. distance error is shown with a green line across the diagram.
The position of the TN, instead, has the following coordinates as in the scenario shown in Fig. 2, then the matrices which
expressed in meters appear in such classic algorithms become ill-conditioned, thus
leading to position estimates which are typically far distant
u = (2.5, 1)T . (24)
from the true position of the TN [12]. At the opposite, the
According to the algorithm outlined in Section II, range PSO algorithm, being a refinement-based algorithm, does not
estimates for each one of the four APs are acquired and used suffer from ill-conditioning related to the geometric relations
to estimate the position of the TN. This procedure is applied among the positions of the APs.
100 times, thus obtaining 100 position estimates {û(j) }100
j=1 for
As done in the first scenario, 100 range estimates from
the TN. Obtained results are then used to evaluate distances each one of the four APs are acquired, but in this case
j=1 between the j-th position estimate and the true
{d(j) }100 the configuration of APs is shown in Fig. 2. The values
position of the TN, according to (21). Such distances are of distances {d(j) }100
j=1 corresponding to the 100 position
used as a performance metrics to investigate the validity of estimates are then evaluated according to (21) and they are
the proposed localization algorithm. Fig. 3 shows the values shown in Fig. 4. Fig. 4 also shows the value of the average
of {d(j) }100
j=1 and also the value of the average localization
localization error davg , which corresponds to 56 cm (green
error davg , which corresponds to 62 cm. Fig. 3 also shows line). Finally, Fig. 4 shows that the distance between the true
that the maximum value of {d(j) }100
j=1 is approximately 1.4 m.
position of the TN and its estimates is always smaller than
This means that the distance between the true position of the 1.7 m, since the maximum value of {d(j) }100 j=1 is 1.68 m.
TN and its estimates is at most 1.4 m, and it is 0.62 m on A comparison between the results shown in Fig. 3 and in
average, which makes the localization sufficiently accurate for Fig. 4 emphasizes that the values of the average localization
many indoor localization purposes. errors davg are close in the two scenarios. The value of davg
Let us now consider the results obtained with the configu- with the configuration of APs shown in Fig. 2, which is the
ration of APs shown in Fig. 2, where all APs are located on less favorable, is 6 cm lower than that obtained with the
the same side of the corridor. In this case, the coordinates of configuration of APs shown in Fig. 1, even if the maximum
the APs expressed in meters are value of {d(j) }100
j=1 is smaller in the first scenario.
s1 = (0, 0)T s2 = (1, 0)T IV. C ONCLUSIONS
T
(25)
s3 = (4, 0) s4 = (5, 0)T .
This paper described and evaluated empirically an
The TN is located in the same point as in the first scenario, and optimization-based localization algorithm which is used to
its coordinates are shown in (24). Note that the configuration provide agents with accurate and timely estimates of their
of APs shown in Fig. 2 is less favorable for localization than position in indoor environments using only ordinary WiFi
that shown in Fig. 1, and many classic localization algorithm infrastructures. The discussed algorithm is implemented in
would fail in estimating the position of the TN. Actually, the the localization add-on module for JADE which provides a
large majority of classic localization algorithm rely on specific framework to host different localization algorithms and to let
geometric considerations concerning the positions of the APs agent developers choose among them to address the specific
and of the TN, and they are based on the solutions of non- needs of applications. In detail, the proposed approach uses
linear systems of equations. If all APs lay on the same line, the communication between the smart device where the agent
70
is running and the APs of the network, which are assumed to the 9th International Wireless Communications & Mobile Computing
be static and in known positions, to acquire estimates of the Conference (IWCMC 2013), Cagliari, Italy, July 2013, pp. 982–987.
[7] S. Monica and G. Ferrari, “UWB-based localization in large indoor
distance between the smart device and each AP. Such estimates scenarios: Optimized placement of anchor nodes,” IEEE Transactions
are used to build a specific optimization problem, which is on Aerospace and Electronic Systems, vol. 51, no. 2, pp. 987–999, April
then solved using PSO. The solution of such an optimization 2015.
[8] S. Monica and F. Bergenti, “Location-aware JADE agents in indoor
problem is an estimate of the position of the smart device scenarios,” in Proceedings of 16th Workshop Dagli Oggetti agli Agenti
which hosts the agent. No dedicated infrastructure is needed (WOA 2015), ser. CEUR Workshop Proceedings, vol. 1382. RWTH
to enable localization, and smart devices are not even requested Aachen, 2015, pp. 103–108.
[9] S. Monica and F. Bergenti, “A comparison of accurate indoor localization
to be connected to one of the available WiFi networks because of static targets via WiFi and UWB ranging,” in Trends in Practical
only standard network discovery messages are used. Applications of Scalable Multi-Agent Systems, the PAAMS Collection,
Possible envisaged applications of the described algorithm 2016, pp. 111–123.
[10] S. Monica and F. Bergenti, “Experimental evaluation of agent-based
include location-aware social games [21], which can be localization of smart appliances,” in Proceedings of the European
adopted, for instance, inside museums and exhibitions to Conference on Multi-Agent Systems (EUMAS), Sevilla, Spain, 2016.
attract the interest of children, or to create personalized [11] Z. Farid, R. Nordin, and M. Ismail, “Recent advances in wireless indoor
localization techniques and system,” Journal of Computer Networks and
itineraries with treasure hunts. Such games can be effectively Communications, vol. 2013, 2013.
implemented using the localization add-on module for JADE [12] S. Monica and G. Ferrari, “Particle swarm optimization for auto-
together with the Agent-based Multi-User Social Environment localization of nodes in wireless sensor networks,” in Proceedings of
the 11th International Conference on Adaptive and Natural Computing
(AMUSE) [22], [23], a recent evolution of JADE that offers Algorithms (ICANNGA 2013), Lausanne, Switzerland, April 2013, pp.
platform-level functionality to help developers in the imple- 456–465.
mentation of social games. Illustrative results presented in [13] S. Monica and G. Ferrari, “Swarm intelligent approaches to auto-
localization of nodes in static UWB networks,” Applied Soft Computing,
the last part of the paper show that the performance of the vol. 25, pp. 426–434, December 2014.
proposed algorithms can be considered sufficiently good for [14] S. Monica and G. Ferrari, “A swarm-based approach to real-time 3D
these types of applications. In addition, location-aware smart indoor localization: Experimental performance analysis,” Applied Soft
Computing, vol. 43, pp. 489–497, June 2016.
emergency applications [24], which were typically intended [15] G. Shen, R. Zetik, and R. S. Thomä, “Performance comparison of TOA
for outdoor environments because they need accurate local- and TDOA based location estimation algorithms in LOS environment,”
ization, could be retargeted to indoor environments with the in Proceedings of the 5th Workshop on Positioning, Navigation and
Communication (WPNC 2008), Hannover, Germany, March 2008, pp.
help of the described algorithm. 71–78.
Future work on the research discussed in this paper involves [16] J. Kennedy and R. Eberhart, “Particle swarm optimization,” in Proceed-
further investigation on the performance of the described ings of the IEEE International Conference on Neural Networks (ICNN),
Perth, Australia, November 1995, pp. 1942–1948.
algorithm in different indoor environments and with different
[17] Y. Shi and R. Eberhart, “A modified particle swarm optimizer,” in
configurations of APs. Moreover, it is of interest to study the Proceedings of the IEEE International Conference on Evolutionary
effects of proper pre-processing of the range estimates in the Computation (ICEC), Washington, DC, July 1999, pp. 69–73.
first phase of the algorithm. Actually, as also suggested by ex- [18] R. Eberhart and J. Kennedy, “A new optimizer using particles swarm
theory,” in Proceedings of the 6th International Symposium on Micro
perimental results shown in previous section, the performance Machine and Human Science (MHS), Nagoya, Japan, October 1995, pp.
of the algorithm could benefit from proper averaging of range 39–43.
estimates or of position estimates. [19] R. Poli, J. Kennedy, and T. Blackwell, “Particle swarm optimization,”
Swarm Intelligence Journal, vol. 1, no. 1, pp. 33–57, June 2007.
R EFERENCES [20] S. Monica and G. Ferrari, “A swarm intelligence approach to 3D
distance-based indoor UWB localization,” in Proceedings of the In-
[1] L. Fichera, F. Messina, G. Pappalardo, and C. Santoro, “A Python ternational Conference on the Applications of Evolutionary Computa-
framework for programming autonomous robots using a declarative tion (EvoApplications 2015), track on Nature-inspired Techniques for
approach,” Science of Computer Programming, vol. 139, pp. 36–55, Communication Networks and other Parallel and Distributed Systems
2017. (EvoCOMNET 2015), Copenaghen, Denmark, April 2015.
[2] S. Poslad, Ubiquitous Computing: Smart Devices, Environments and [21] F. Bergenti and S. Monica, “Location-aware social gaming with
Interactions. Wiley, 2009. AMUSE,” in Advances in Practical Applications of Scalable Multi-
[3] F. Bergenti, G. Caire, and D. Gotta, “Agents on the move: JADE for agent Systems. The PAAMS Collection: 14th International Conference,
Android devices,” in Proceedings of the 15th Workshop Dagli Oggetti PAAMS 2016, Y. Demazeau, T. Ito, J. Bajo, and M. J. Escalona, Eds.
agli Agenti (WOA 2014), ser. CEUR Workshop Proceedings, vol. 1260. Springer International Publishing, 2016, pp. 36–47.
RWTH Aachen, 2014. [22] F. Bergenti, G. Caire, and D. Gotta, “An overview of the AMUSE social
[4] S. Gezici and H. V. Poor, “Position estimation via Ultra-Wide-Band gaming platform,” in Proceedings of the 14th Workshop Dagli Oggetti
signals,” Proceedings of the IEEE, vol. 97, no. 2, pp. 386–403, February agli Agenti (WOA 2013), ser. CEUR Workshop Proceedings, vol. 1099.
2009. RWTH Aachen, 2013.
[5] S. Monica and G. Ferrari, “Impact of the number of beacons in [23] F. Bergenti, G. Caire, and D. Gotta, “Agent-based social gaming with
PSO-based auto-localization in UWB networks,” in Proceedings of the AMUSE,” in Procs. 5th Int’l Conf. Ambient Systems, Networks and
International Conference on the Applications of Evolutionary Compu- Technologies (ANT 2014) and 4th Int’l Conf. Sustainable Energy
tation (EvoApplications 2013), track on Nature-inspired Techniques for Information Technology (SEIT 2014), ser. Procedia Computer Science,
Communication Networks and other Parallel and Distributed Systems vol. 32, 2014, pp. 914–919.
(EvoCOMNET 2013), Vienna, Austria, April 2013, pp. 42–51. [24] A. Poggi and F. Bergenti, “Developing smart emergency applications
[6] S. Monica and G. Ferrari, “Optimized anchors placement: An analyt- with multi-agent systems,” International Journal of E-Health and Med-
ical approach in UWB-based TDOA localization,” in Proceedings of ical Communications, vol. 1, no. 4, pp. 1–13, 2010.