Kim et al.

Big Data Analytics (2018) 3:4

https://doi.org/10.1186/s41044-018-0031-2
Big Data Analytics

RESEARCH Open Access

A scalable deep neural network

architecture for multi-building and multi-floor
indoor localization based on Wi-Fi
fingerprinting
Kyeong Soo Kim1,2* , Sanghyuk Lee1,2 and Kaizhu Huang1
An earlier version of this paper was presented in part at ICUMT/FOAN 2017, Munich, Germany, November 2017.

*Correspondence:
[email protected] Abstract
1
Department of Electrical and Background: One of the key technologies for future large-scale location-aware
Electronic Engineering, XJTLU, 111
Ren’ai Road, 215123 Suzhou, China
services covering a complex of multi-story buildings is a scalable indoor localization
2
Centre for Smart Grid and technique. In this paper, we report the current status of our investigation on the use of
Information Convergence, XJTLU, deep neural networks (DNNs) for the scalable building/floor classification and
111 Ren’ai Road, 215123 Suzhou,
China
floor-level position estimation based on Wi-Fi fingerprinting. Exploiting the hierarchical
nature of the building/floor estimation and floor-level coordinates estimation of a
location, we propose a new DNN architecture consisting of a stacked autoencoder for
the reduction of feature space dimension and a feed-forward classifier for multi-label
classification of building/floor/location, on which the multi-building and multi-floor
indoor localization system based on Wi-Fi fingerprinting is built.
Results: We evaluate the performance of building/floor estimation and floor-level
coordinates estimation of a given location using the UJIIndoorLoc dataset covering
three buildings with four or five floors in the Jaume I University (UJI) campus, Spain.
Experimental results demonstrate the feasibility of the proposed DNN-based indoor
localization system, which can provide near state-of-the-art performance using a single
DNN.
Conclusions: The proposed scalable DNN architecture for multi-building and
multi-floor indoor localization based on Wi-Fi fingerprinting can achieve near
state-of-the-art performance with just a single DNN and enables the implementation
with lower complexity and energy consumption at mobile devices.
Keywords: Multi-building and multi-floor indoor localization, Wi-Fi fingerprinting,
Deep learning, Neural networks, Multi-label classification

Background
Location fingerprinting using received signal strengths (RSSs) from wireless network
infrastructure is one of the most popular and promising technologies for localization in
an indoor environment, where there is no line-of-sight signal from the global position-
ing system (GPS) available [1]: For example, a vector of pairs of a medium access control
(MAC) address and an RSS for a Wi-Fi access point (AP) measured at a location can
© The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://
creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Kim et al. Big Data Analytics (2018) 3:4 Page 2 of 17

be its location fingerprint, which is illustrated in Fig. 1. A location of a user/device then

can be estimated by finding the closest match between its RSS measurement and the
fingerprints of known locations in a database [2] as shown in Fig. 2. Note that the loca-
tion fingerprinting technique does not require the installation of any new infrastructure
or the modification of existing devices but that it is just based on the existing wireless
infrastructure, which is its major advantage over alternative techniques like triangulation
based on time of arrival (TOA) requiring precise synchronization among all transmitters
and receivers in the system and non-standard timestamp labeling for the measurement of
distances between a target and reference points [3].
In this paper, we report the current status of our investigation on the use of deep
neural networks (DNNs) in multi-building and multi-floor indoor localization based
on Wi-Fi fingerprinting. We propose a new DNN architecture consisting of a stacked
autoencoder (SAE) [4] for the reduction of feature space dimension and a feed-forward
multi-label classifier [5, 6] for the scalable building/floor classification and floor-level
location estimation and evaluate its performance using the UJIIndoorLoc dataset [7].

Multi-building and multi-flow indoor localization

When the indoor localization is to cover a large building complex — e.g., a big shop-
ping mall or a university campus — where there are lots of multi-story buildings under
the same management, the scalability of fingerprinting techniques becomes an important
issue. The current state-of-the-art Wi-Fi fingerprinting techniques assume a hierarchical
approach to the indoor localization, where the building, floor, and position (e.g., a label or
coordinates) of a location are estimated in a hierarchical and sequential way using a dif-
ferent algorithm tailored for each task. In [8], for instance, building estimation is done as
follows: Given the AP with the strongest RSS in a measured fingerprint, we first build a
subset of fingerprints where the same AP has the strongest RSS; then, we count the num-
ber of fingerprints associated to each building and set the estimated building to be the

Fig. 1 A location fingerprint example based on Wi-Fi received signal strengths (RSSs)
Kim et al. Big Data Analytics (2018) 3:4 Page 3 of 17

Fig. 2 An overview of Wi-Fi fingerprinting procedure

most frequent one from the counting. Similar procedures are also proposed to estimate
a floor inside the building. For the estimation of the coordinates of the location, we first
build a subset of fingerprints belonging to the building and the floor estimated from the
previous procedures. Then, take multiple fingerprints from the subset most similar to the
measured one, and compute the centroid of the coordinates of the selected fingerprints
as the estimated coordinates of the given location. According to the results in [8], the best
building and floor hit rates achieved for the UJIIndoorLoc dataset [7] are 100% and 94%,
respectively, and the mean error in coordinates estimation is 6.20m1 .

Wi-Fi fingerprinting based on deep neural networks

One of the major challenges in Wi-Fi fingerprinting is how to deal with the random fluctu-
ation of a signal, the noise from multi-path effects, and the device & position dependency
in RSS measurements. Unlike traditional solutions relying on complex filtering and time-
consuming parameter tuning specific to given conditions, the popular DNNs can provide
attractive solutions to Wi-Fi fingerprinting due to their less parameter tuning and adapt-
ability to a wider range of conditions with standard architectures and training algorithms
[9–11]: In [9], a four-layer DNN generates a coarse positioning estimate, which, in turn,
is refined to produce a final position estimate by a hidden Markov model (HMM)-based
fine localizer. The performance of the proposed indoor localization system is evaluated in
both indoor and outdoor environments which are divided into hundreds of square grids.
In [10], the authors investigate the application of deep belief networks (DBNs) with two
different types of Restricted Boltzmann Machines for indoor localization and evaluate
the performance of their approaches using data from simulation in heterogeneous mobile
radio networks using ray tracing techniques. In both cases, the authors focus only on the
localization in a single plane and do not consider the hierarchical nature of multi-building
and multi-floor indoor localization. In [11], on the other hand, a DNN consisting of an
SAE and a feed-forward multi-class classifier is used for building/floor classification. This
Kim et al. Big Data Analytics (2018) 3:4 Page 4 of 17

work, too, does not take into account the hierarchical nature of building/floor classifica-
tion, because the classification is done over flattened, one-dimensional labels of combined
building and floor identifiers. Also, the floor-level location estimation is not considered
at all. In this regard, to the best of our knowledge, the work presented in this paper is the
first to apply DNNs for multi-building and multi-floor indoor localization, exploiting its
hierarchical nature in classification.

A scalable DNN architecture for multi-building and multi-floor indoor localization based on
Wi-Fi fingerprinting
Location awareness is one of enabling technologies for future smart and green cities;
understanding where people spend their times and how they interact with environments
is critical to realizing the vision of smart and green cities [12]. One of the key technologies
for future large-scale location-aware services covering a complex of multi-story buildings
— e.g., a big shopping mall and a university campus — is a scalable indoor localization
technique. Regarding the scalability of the indoor localization, consider the evolution of
the Xi’an Jiaotong-Liverpool University (XJTLU) campus in Suzhou, China, where the
authors are currently working: As shown in Fig. 3a, the XJTLU started with just one build-
ing in 2006. As of this writing, the XJTLU has two campuses, which are shown in Fig. 3b,
and the number of buildings over two campuses has increased to around 20; this number

Fig. 3 XJTLU campus in (a) 2006 and (b) 2017

Kim et al. Big Data Analytics (2018) 3:4 Page 5 of 17

is still increasing as more buildings and sports facilities are being constructed. Consider-
ing all the floors within each building and the locations on each floor, the total number
of distinct locations (e.g., offices, lecture rooms, and labs) is already on the order of thou-
sands. If we adopt a grid-based representation of the localization area as in [9], the total
number of locations would be even greater. The indoor localization system to cover such
a large building complex, therefore, must be scalable.
Figures 4 and 5 show two alternative system architectures for large-scale DNN-based
multi-building and multi-floor indoor localization. In the hierarchical architecture shown
in Fig. 4, the task of building/floor/location classification is separated into multiple sub-
tasks dedicated to the classification at each level of building, floor, and location. This
architecture directly corresponds to the state-of-the-art hierarchical Wi-Fi fingerprinting
methods (e.g., [8]), where DNNs replace traditional techniques for building, floor, and
location estimation. Compared to the methods based on traditional techniques, a major
disadvantage in this hierarchical DNN architecture is that the DNNs in the floor and
the location levels of the system need to be trained separately with multiple sub-datasets
derived from a common dataset (i.e., building-specific datasets for DNNs for floor estima-
tion and building-floor-specific datasets for DNNs for location estimation), which poses
significant challenges on the management of location fingerprint databases as well as the
training of possibly a large number of DNNs. In this paper, therefore, we focus on the
integrated architecture shown in Fig. 5 where a single DNN handles the classification of
building, floor, and location in an integrated way with a common dataset.

Fig. 4 A hierarchical system architecture for large-scale DNN-based multi-building and multi-floor indoor
localization
Kim et al. Big Data Analytics (2018) 3:4 Page 6 of 17

Fig. 5 An integrated system architecture for large-scale DNN-based multi-building and multi-floor indoor
localization

Figure 6 shows an SAE used for the reduction of feature space dimension; after train-
ing with RSSs as both input and output data as shown in the figure, only the gray-colored
nodes are used as an encoder for feature space dimension reduction. Figure 7 shows a
DNN architecture for the combined estimation of building, floor, and location consisting
of the encoder part of the SAE and a feed-forward classifier for multi-class classifica-
tion with flattened building-floor-location labels, which is a straightforward extension of
the DNN system proposed in [11] for building/floor classification. This DNN architec-
ture based on multi-class classification with flattened labels, however, has the scalability
issue that the number of output nodes is equal to the number of locations over the build-
ing complex: In case of the UJIIndoorLoc dataset, the number of distinct locations (i.e.,
also called reference points in [7]) over three buildings with four or five floors is 933.
It also does not take into account the hierarchical nature of the building/floor/location

Fig. 6 A stacked autoencoder (SAE) for the reduction of feature space dimension
Kim et al. Big Data Analytics (2018) 3:4 Page 7 of 17

Fig. 7 A DNN architecture for the combined estimation of building, floor, and location which consists of the
encoder part of the SAE and a feed-forward classifier for multi-class classification with flattened
building-floor-location labels as in [11]

classification problem due to its calculating the loss and the accuracy over flattened build-
ing/floor/location labels2 ; the misclassification of building, floor, or location has equal loss
during the training phase. To reflect the hierarchical nature of the building/floor/location
classification in a DNN classifier, one can use a hierarchical loss function — e.g., a
loss function with different weights for building, floor, and location — with the exist-
ing multi-class classifier and flattened labels. Because the hierarchical loss function for
flattened labels is quite complicated and does not provide a closed-form gradient func-
tion, however, training the DNN with the usual backpropagation procedure could be
challenging.
To address the scalability issue of the DNN classifier based on multi-class classification
and take into account the hierarchical nature of the building/floor/location classification,
we propose a scalable DNN architecture based on multi-label classification3 shown in
Fig. 8. The building/floor/location classification with the proposed architecture is done as

Fig. 8 A DNN architecture for the scalable building/floor classification and floor-level coordinates estimation
based on an SAE for the reduction of feature space dimension and a feed-forward classifier for multi-label
classification
Kim et al. Big Data Analytics (2018) 3:4 Page 8 of 17

follows: First, building, floor, and location identifiers are mapped to sequential numbers,
the latter two of which are meaningful only in combination with higher-level numbers;
those numbers are one-hot encoded independently and combined together into a vec-
tor as a categorical variable for multi-label classification as illustrated in Table 1. Then,
the output vector from the multi-label classifier is split into a building, a floor, and a
location vector by indexes as shown in Fig. 8. Finally, we estimate the building and the
floor of a location as the index of a maximum value of the corresponding vector through
the arg max function. For the estimation of the location coordinates, we select κ largest
 
elements from the location vector (i.e., L = L1 , . . . , Lmax(...,NL (i.j)) in Fig. 8), filter out
the elements whose values are less than σ × max(L) (σ ∈ [0, 1]), and calculate the esti-
mated coordinates of the location as either a normal or a weighted (with the values
of the elements as weights) centroid of the remaining elements as described in detail
in Fig. 9.
Note that there are two design parameters — i.e., κ and σ — in the location coor-
dinates estimation procedure, the rationale of which is illustrated in Fig. 10: If we use
only κ as a design parameter as in [8] and sets its value to 5 in Fig. 10a, we can include
the reference points quite close to the new location (i.e., those inside the dotted cir-
cle) in the estimation procedure and can generate good estimation. In Fig. 10b, however,
the same value of κ could result in poor estimation because the reference points 4 and
5 have to be considered during the estimation. With κ = 3, on the other hand, we can
expect good estimation with Fig. 10b but not with Fig. 10a this time. If we can use
both κ and σ as design parameters, however, we can include good reference points by
properly setting σ for a threshold value. The actual effects of these design parameters
on the location coordinates estimation are investigated in “Results” and “Discussion”
sections.

Table 1 Label formation example for the multi-label classification of building, floor, and location
with two multi-story buildings
Building Floor Location Sequential coding One-hot coding
A 1st Lecture Theater A 0,0,0 01|0001|0001
Lecture Theater B 0,0,1 01|0001|0010

2nd Lab 1 0,1,0 01|0010|0001

Lab 2 0,1,1 01|0010|0010

3rd A301 0,2,0 01|0100|0001

A302 0,2,1 01|0100|0010
A303 0,2,2 01|0100|0100

B 1st Common Room 1,0,0 10|0001|0001

Printing Room 1,0,1 10|0001|0010

2nd Conference Room 1 1,1,0 10|0010|0001

Conference Room 2 1,1,1 10|0010|0010

3rd B301 1,2,0 10|0100|0001

B302 1,2,1 10|0100|0010
B303 1,2,2 10|0100|0100
B304 1,2,3 10|0100|1000

4th Gym 1,3,0 10|1000|0001

Kim et al. Big Data Analytics (2018) 3:4 Page 9 of 17

Fig. 9 Procedure for the estimation of location coordinates

As for the scalability of the proposed DNN architecture, the number of output nodes
becomes much smaller than that of the DNN architecture based on multi-class classifica-
tion: The number of output nodes for multi-label building/floor/location classification is
given by

NB + max (NF (1), . . . , NF (NB )) + max (NL (1, 1), . . . , NL (NB , NF (NB ))) , (1)

where NB , NF (i), and NL (j, k) are the number of buildings in the complex, the number of
floors in the ith building (i = 1, . . . , NB ), and the number of locations on the kth floor of
the jth building (j = 1, . . . , NB ; k = 1, . . . , NF (j)), respectively. Note that for multi-class
building/floor/location classification, the number becomes

 
NB N F (i)
NL (i, j). (2)
i=1 j=1

According to (1), the number of output nodes of the proposed DNN architecture for
the publicly available UJIIndoorLoc dataset at the University of California, Irvine (UCI),
Machine Learning Repository4 is given by 118 (i.e., the sum of the number of buildings
(3), the maximum of the numbers of floors of the buildings (5), and the maximum of the
numbers of locations5 on the floors (110)), which is smaller than the number of output
nodes of the DNN architecture based on multi-class classification (i.e., 9056 ). Note that
the difference could be much larger if the UJIIndoorLoc dataset covers all the buildings
on the Jaume I University (UJI) campus where the data were collected.
Kim et al. Big Data Analytics (2018) 3:4 Page 10 of 17

Fig. 10 Location coordinates estimation examples (a) with several reference points centered around a new
location and (b) with only a few reference points around it, where × denotes a new location whose
coordinates are to be estimated and  denotes a reference point whose coordinates are stored in the
fingerprint database. The numbers over the reference points indicate their closeness to the new location
with 1 being the closest

Also, due to the clear mapping between building, floor, and location identifiers and its
corresponding one-hot-encoded categorical variable for the DNN-based multi-label clas-
sifier, it is easy to carry out different processing for parts of DNN outputs specifically for
building, floor, and location as illustrated in Fig. 8. Especially, the use of multiple elements
in estimating location coordinates as described in Fig. 9 is a huge advantage in terms of
computational complexity because trained DNNs can generate multi-dimensional output
values in parallel; in traditional approaches, on the other hand, selecting nearest locations
based on Euclidean distances are complex and time consuming. This flexibility in han-
dling DNN outputs also makes it easy to apply different weights to the cost of building,
floor, and location classification error during the training phase.

Methods
We carry out experiments using the UJIIndoorLoc dataset [7] to evaluate the perfor-
mance of the proposed DNN-based multi-building and multi-floor indoor localization
system. DNN models are implemented based on Keras [13] and TensorFlow [14], and
their source code is available online [15]. We focus on the effects of the number of largest
Kim et al. Big Data Analytics (2018) 3:4 Page 11 of 17

elements from the output location vector (i.e., κ) and the scaling factor for a threshold
(i.e., σ ) in the location coordinates estimation procedure described in “A scalable DNN
architecture for multi-building and multi-floor indoor localization based on Wi-Fi
fingerprinting” section. Table 2 summarizes DNN parameter values for the experiments,
which are chosen experimentally and used throughout the experiments.
As indicated in [11], the publicly available UJIIndoorLoc dataset includes training
and validation data, but not testing data which were provided only to the competitors
at the Evaluating Ambient Assisted Living (EvAAL) competition at the International
Conference on Indoor Positioning and Indoor Navigation (IPIN) 2015 [8]. Also, unlike the
training data, the validation data do not include location information (i.e., SpaceID and
RELATIVEPOSITION fields) because the measurements were taken at arbitrary points as
would happen in a real localization system. In this regard, we split the training data into
new training and validation data with the ratio of 70:30 for DNN training and validation
with building/floor/location labels for both. During the evaluation phase, the output from
the trained DNN are post-processed as described in “A scalable DNN architecture for
multi-building and multi-floor indoor localization based on Wi-Fi fingerprinting” section
and compared with the building, floor, and coordinates of a given location. In this way,
we can compare out results of multi-building and multi-floor indoor localization with the
baseline and the best results from [7] and [8], respectively.

Results
Table 3 summarizes our experimental results, which show the effects of the number of
largest elements from the output location vector (κ) and the scaling factor for a threshold
(σ ) on the performance of multi-building and multi-floor indoor localization.

Discussion
In general, σ in the range of 0.1–0.3 produces the best localization performance for κ≤8;
once κ becomes larger than 8, however, higher values of σ (i.e., 0.4 for κ = 9 and 0.5
for κ = 10) generate better performance. Considering the coordinates location estimation
examples shown in Fig. 10 with their explanations in “A scalable DNN architecture for
multi-building and multi-floor indoor localization based on Wi-Fi fingerprinting” section
regarding the use of two design parameters, we can explain these results as follows:

Table 2 Parameter Values for Scalable DNN-Based Indoor Localization

DNN parameter Value
Ratio of training data to overall data 0.90
Number of Epochs 20
Batch size 10
SAE hidden layers 256-128-256
SAE activation Rectified Linear (ReLU)
SAE Optimizer ADAM [25]
SAE loss Mean Squared Error (MSE)
Classifier hidden layers 64-128
Classifier activation ReLU
Classifier optimizer ADAM
Classifier loss Binary Crossentropy
Classifier dropout rate 0.20
Kim et al. Big Data Analytics (2018) 3:4 Page 12 of 17

Table 3 Effects of the number of largest elements from the output location vector (κ) and the scaling
factor for a threshold (σ ) on the Performance of multi-building and multi-floor indoor localization
Positioning error [m]
κ σ Building hit rate [%] Floor hit rate [%] Success rate [%]
Centroid Weighted centroid
1 N/A∗ 99.82 91.90 91.81 11.40 11.40

2 0.0 99.37 92.44 91.81 10.62 10.54

0.1 100.00 91.81 91.81 10.40 10.33
0.2 99.82 92.62 92.44 9.74 9.66
0.3 99.64 91.99 91.81 9.78 9.71
0.4 99.73 91.54 91.45 10.29 10.21
0.5 100.00 90.01 90.01 10.16 10.09

3 0.0 99.73 91.54 91.36 10.14 9.79

0.1 99.91 90.91 90.82 9.92 9.76
0.2 98.83 90.91 90.28 9.98 9.80
0.3 99.55 92.08 91.90 10.13 10.01
0.4 99.91 91.99 91.99 10.63 10.47
0.5 99.82 90.37 90.37 9.94 9.89

4 0.0 99.82 90.91 90.91 10.27 9.66

0.1 99.37 91.99 91.63 10.37 9.92
0.2 99.64 92.08 91.90 10.26 10.09
0.3 99.82 91.45 91.36 10.24 10.16
0.4 99.91 92.26 92.17 10.35 10.23
0.5 99.82 91.27 91.18 10.10 10.07

5 0.0 99.91 91.36 91.27 11.29 10.36

0.1 99.91 91.63 91.63 9.90 9.62
0.2 99.91 90.73 90.73 9.89 9.57
0.3 99.82 90.91 90.82 10.27 9.99
0.4 99.73 92.17 92.08 10.17 10.01
0.5 99.82 92.98 92.89 10.59 10.54

6 0.0 99.82 91.90 91.72 10.84 9.71

0.1 99.64 92.08 91.81 10.35 9.86
0.2 100.00 91.99 91.99 9.85 9.56
0.3 99.82 92.80 92.80 10.49 10.22
0.4 99.37 91.09 91.00 10.32 10.17
0.5 99.64 90.91 90.64 9.55 9.52

7 0.0 99.82 89.29 89.29 11.74 10.22

0.1 99.82 90.19 90.01 10.43 9.82
0.2 99.91 91.45 91.45 10.00 9.55
0.3 99.91 91.63 91.54 9.75 9.53
0.4 99.64 90.46 90.19 10.42 10.28
0.5 99.55 91.45 91.36 9.83 9.73

8 0.0 99.91 90.19 90.10 11.32 9.27

0.1 100.00 91.27 91.27 10.62 10.14
0.2 99.82 91.27 91.18 9.76 9.29
0.3 99.82 90.55 90.37 9.95 9.82
0.4 99.91 90.37 90.28 10.21 10.14
0.5 99.91 90.55 90.55 9.86 9.79
Kim et al. Big Data Analytics (2018) 3:4 Page 13 of 17

Table 3 Effects of the number of largest elements from the output location vector (κ) and the
scaling factor for a threshold (σ ) on the Performance of multi-building and multi-floor indoor
localization (Continued)
Positioning error [m]
κ σ Building hit rate [%] Floor hit rate [%] Success rate [%]
Centroid Weighted centroid
9 0.0 99.82 91.00 90.91 12.75 9.76
0.1 99.82 90.28 90.19 10.17 9.55
0.2 99.82 90.73 90.64 10.05 9.76
0.3 99.82 91.45 91.36 10.36 10.23
0.4 100.00 91.81 91.81 9.86 9.78
0.5 99.91 92.80 92.71 10.26 10.23
10 0.0 99.91 90.91 90.82 13.21 9.75
0.1 99.82 92.71 92.62 10.37 9.85
0.2 99.91 91.72 91.72 10.33 10.04
0.3 100.00 90.82 90.82 10.47 10.24
0.4 100.00 91.09 91.09 10.10 9.96
0.5 100.00 91.90 91.90 9.96 9.91
*
N/A = not applicable; when κ = 1, the value of σ does not affect the selection of locations (i.e., reference points) included in the
coordinates estimation

With a larger value of κ (i.e., 9 and 10), there could be a higher chance of includ-
ing reference points relatively far from the given location as shown in Fig. 10b. In
such a case, a tighter threshold (i.e., a larger value of σ 7 ) can filter out those reference
points.
According to the results shown in Table 3, collectively the best results are achieved
when κ = 8 and σ = 0.2. These results from the proposed DNN-based multi-building
and multi-floor indoor localization system — i.e., 99.82% for building hit rate,
91.27% for floor hit rate, 91.18% for success rate and 9.29m for positioning error
— are favorably comparable to the baseline results — i.e., 89.92% for success rate
and 7.9m for positioning error — from [7] which are based on the distance-based
k-Nearest Neighbors (kNN) algorithm [16]. As discussed in “A scalable DNN architecture
for multi-building and multi-floor indoor localization based on Wi-Fi fingerprinting”
section, even though a direct comparison with the results from the EvAAL/IPIN 2015
competition is not possible due to the lack of testing samples in the public version
of the UJIIndoorLoc dataset and a slightly different way of calculating the position-
ing error, our results are also comparable to the competition results summarized in
Table 4.
Note that the results presented in this section are not optimized with DNN parameters,
including the number of hidden layers and the number of nodes at each layer; we inves-
tigated the feasibility of the combined building/floor/location estimation using a single

Table 4 Best results from the four teams at the EvAAL/IPIN 2015 competition [8]
MOSAIC HFTS RTLSUM ICSL
Building hit rate [%] 98.65 100 100 100
Floor hit rate [%] 93.86 96.25 93.74 86.93
Positioning error (Mean) [m] 11.64 8.49 6.20 7.67
Positioning error (Median) [m] 6.7 7.0 4.6 5.9
Kim et al. Big Data Analytics (2018) 3:4 Page 14 of 17

DNN based on multi-label classification framework with focus on the effects of the num-
ber of largest elements from the output location vector (κ) and the scaling factor for a
threshold (σ ) in the location coordinates estimation. This leaves much room for further
optimization of the performance.

Open issues and areas for further research

There are several open issues raised by the current work and related areas for further
research in multi-building and multi-floor indoor localization based on DNNs and Wi-Fi
fingerprinting.
First, in the current work we focus on the estimation of the position of a static
user/device based on location fingerprints consisting of MAC addresses and RSSs of
Wi-Fi access points only, which could be extended to the case of a moving user/device.
In fact, during the more recent IPIN competitions in 2016 and 2017 [17–19], the focuses
have been shifted to more challenging trajectory estimation based on continuously-
recorded data from multiple sensors of advanced smartphones — including inertial data,
magnetic field, GPS, and pressure in addition to Wi-Fi RSS — and additional informa-
tion like floorplan maps and map-based reference trajectories. Note that, while the static
indoor localization based on a minimal set of information (i.e., Wi-Fi RSS) could be con-
sidered as the limit of the current work, it is also its strength in that this type of indoor
localization makes it easy to build and maintain the fingerprint database and can be
applicable to a variety of mobile devices including smartphones, smartwatches, tablets,
laptops, wireless sensor nodes, and even Internet of Things (IoTs) devices; the location
fingerprints based on Wi-Fi RSS are a common denominator of the information available
from any Wi-Fi-capable devices.
Second, there is no direct match between the cost function used in DNN train-
ing/validation and the actual performance in the final evaluation for building/floor
detection and the location coordinates estimation due to the additional processing of
DNN output (i.e., the use of arg max function for the former and the rather compli-
cated processing with multiple elements of the location vector for the latter). To fully take
into account the actual performance of building/floor detection and location coordinates
estimation during training/validation, we may consider heuristics like evolutionary algo-
rithms (e.g., genetic algorithm (GA) [20] and particle swarm optimization (PSO) [21]),
simulated annealing [22], and quantum annealing [23] for training DNN weights; due to
its many tradeoffs between complexity and flexibility resulting from the use of heuris-
tics in DNN weight training, this approach could be an interesting topic for long-term
research.
Third, the issue of building and maintenance of a large-scale fingerprint database
should be further investigated, too. For instance, it is a real challenge to gather
the location fingerprints covering the vast area of a large building complex either
directly from researchers and volunteers or through outsourcing to designated pro-
fessionals or an organization. In this regard, crowdsourcing to a general public (e.g.,
students and staff members of a university or shoppers and visitors of a shopping
mall) could be a viable solution [24]. Note that, however, crowdsourcing could raise
other issues like malicious users intentionally submitting wrong data (e.g., RSSs with
wrong location labels) in order to corrupt the fingerprint database. Also, how to
update the fingerprint database with new measurement data, which could have different
Kim et al. Big Data Analytics (2018) 3:4 Page 15 of 17

statistics, is another important issue. In the current work based on the UJIIndoorLoc
dataset, we already found that the statistics of its training and validation data, which
were generated at different times and by different Android apps, show significant
differences.

Conclusions
In this paper we have proposed a new scalable DNN architecture for multi-building and
multi-floor indoor localization based on Wi-Fi fingerprinting, which can cover a large-
scale complex of many multi-story buildings under the same management. The proposed
DNN architecture consists of an SAE for the reduction of feature space dimension and a
feed-forward classifier for multi-label classification of building/floor/location. Reformu-
lating the problem of building/floor/location classification based on the framework of
multi-label classification, we can achieve better scalability (i.e., greatly reducing the num-
ber of DNN output nodes) and better exploit the hierarchical nature of the building/floor
estimation and the floor-level location coordinates estimation through systematic label
formation (i.e., providing straightforward mapping between a DNN categorical variable
and building, floor & location vectors) compared to existing DNN architectures based on
multi-class classification.
The experimental results using the UJIIndoorLoc dataset clearly demonstrate the fea-
sibility of the proposed DNN-based multi-building and multi-floor indoor localization
system, which can provide near state-of-the-art performance using a single DNN in an
integrated way. Combined with the unique advantage of a DNN-based indoor localization
system that, once trained, it does not need the fingerprint database any longer but carries
the necessary information for localization in DNN weights, the scalable DNN architec-
ture proposed in this paper could open a door for a future secure and energy-efficient
indoor localization solution exclusively running on mobile devices without exchanging
any data with the server.

Endnotes
1
The mean error takes into account the building and floor estimation penalties; refer
to [8, Eq. (2)] for details.
2
For example, we can form a flattened label “Bi -Fi,j -Li,j,k ” by combining a building, a
floor, and a location label, where Bi , Fi,j , and Li,j,k denote the ith building, jth floor of the
building, and kth location on the floor, respectively.
3
In multi-class classification (also called single-label classification), an instance is asso-
ciated with only a single label from a set of disjoint labels; in multi-label classification, on
the other hand, an instance can be associated with multiple labels [5].
4
https://archive.ics.uci.edu/ml/datasets/ujiindoorloc.
5
In the UJIIndoorLoc dataset, the position of a location is uniquely determined by four
identifiers, i.e., BuildingID, Floor, SpaceID, and RELATIVEPOSITION. For convenience,
we combine the SpaceID and the RELATIVEPOSITION into one and mention it as loca-
tion throughout the paper so that the three identifiers for building, floor, and location
uniquely determine the position of a location.
6
There are slight differences between the statistics of the UJIIndoorLoc dataset
described in [7] and those of the publicly available dataset at the UCI Machine Learning
Repository.
Kim et al. Big Data Analytics (2018) 3:4 Page 16 of 17

7
When σ = 0, there is no filtering (i.e., including all κ reference points); when σ = 1,
only the reference point with the largest value is considered during the location coordi-
nates estimation.
Acknowledgements
Not applicable.

Funding
This work was supported in part by Xi’an Jiaotong-Liverpool University (XJTLU) Research Development Fund (under
Grant RDF-14-01-25), Summer Undergraduate Research Fellowships programme (under Grant SURF-201739), Research
Institute for Smart and Green Cities Seed Grant Programme 2016-2017 (under Grant RISGC-2017-4), Centre for Smart Grid
and Information Convergence, National Natural Science Foundation of China (NSFC) (under Grant 61473236), Natural
Science Fund for Colleges and Universities in Jiangsu Province (under Grant 17KJD520010), and Suzhou Science and
Technology Program (under Grant SYG201712, SZS201613).

Availability of data and materials

All the data and source code for DNN models are available in the GitHub repository for research on indoor localization
based on wireless fingerprinting techniques at XJTLU [15], https://github.com/kyeongsoo/indoor_localization.

Authors’ contributions
KSK, SL, and KH initiated a seed project on the subject of scalable indoor localization based on DNNs and Wi-Fi
fingerprinting, which this manuscript is based on. KSK proposed DNN architectures, carried out numerical experiments,
and drafted the manuscript, which SL and KH checked and clarified. All authors read and approved the final manuscript.

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests
The authors declare that they have no competing interests.

Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Received: 2 January 2018 Accepted: 3 April 2018

References
1. He S, Chan S-HG. Wi-Fi fingerprint-based indoor positioning: Recent advances and comparisons. IEEE Commun Surv
Tuts. 2016;18(1):466–90.
2. Bahl P, Padmanabhan VN. RADAR: An in-building RF-based user location and tracking system. In: Proc. 2000 IEEE
INFOCOM. vol 2. 2000. p. 775–84. https://doi.org/10.1109/INFCOM.2000.832252.
3. Liu H, Darabi H, Banerjee P, Liu J. Survey of wireless indoor positioning techniques and systems. IEEE Trans. Syst.
Man Cybern. C. 2007;37(6):1067–80.
4. Vincent P, Larochelle H, Lajoie I, Bengio Y, Manzagol P-A. Stacked denoising autoencoders: Learning useful
representations in a deep network with a local denoising criterion. J Mach Learn Res. 2010;11:3371–408.
5. Tsoumakas G, Katakis I. Multi-label classification: An overview. Int J Data Warehousing Mining. 2007;3(3):1–13.
https://doi.org/10.4018/jdwm.2007070101.
6. Herrera F, Charte F, Rivera AJ, del Jesus MJ. Multilabel classification. In: Multilabel Classification: Problem Analysis,
Metrics and Techniques, Chap. 2. Switzerland: Springer; 2016. p. 17–31.
7. Torres-Sospedra J, Montoliu R, Martínez-Usó A, Avariento JP, Arnau TJ, Benedito-Bordonau M, Huerta J.
UJIIndoorLoc: A new multi-building and multi-floor database for WLAN fingerprint-based indoor localization
problems. In: Proc. International Conference on Indoor Positioning and Indoor Navigation (IPIN). Piscataway: IEEE;
2014. p. 261–70.
8. Moreira A, Nicolau MJ, Meneses F, Costa A. Wi-Fi fingerprinting in the real world – RTLSUM at the EvAAL
competition. In: Proc. International Conference on Indoor Positioning and Indoor Navigation (IPIN). Piscataway: IEEE;
2015. p. 1–10.
9. Zhang W, Liu K, Zhang W, Zhang Y, Gu J. Deep neural networks for wireless localization in indoor and outdoor
environments. Neurocomputing. 2016;194:279–87.
10. Félix G, Siller M, Álvarez EN. A fingerprinting indoor localization algorithm based deep learning. In: Proc. ICUFN
2016. Vienna; 2016. p. 1006–11.
11. Nowicki M, Wietrzykowski J. Low-effort place recognition with wifi fingerprints using deep learning. ArXiv e-prints.
2017. arXiv:1611.02049v2 [cs.RO].
12. Benevolo C, Dameri RP, D’Auria B. Smart mobility in smart city: Action taxonomy, ICT intensity and public benefits.
In: Torre T, Braccini AM, Spinelli R, editors. Empowering Organizations: Lecture Notes in Information Systems and
Organisation, 1st edn, Lecture notes in information systems and organisation. Chap. 2. Switzerland: Springer; 2016.
p. 13–28.
Kim et al. Big Data Analytics (2018) 3:4 Page 17 of 17

13. Chollet F, et al. Keras. https://github.com/keras-team/keras/. Accessed 12 Apr 2018.

14. TensorFlowTM . https://www.tensorflow.org/. Accessed 29 Dec 2017.
15. GitHub Repository for Research on Indoor Localization Based on Wireless Fingerprinting Techniques at XJTLU.
https://github.com/kyeongsoo/indoor_localization. Accessed 29 Dec 2017.
16. Cover TM, Hart PE. Nearest neighbor pattern classification. IEEE Trans Inf Theory. 1967;13(1):21–7.
17. Torres-Sospedra J, Jiménez AR, Knauth S, Moreira A, Beer Y, Fetzer T, Ta V-C, Montoliu R, Seco F,
Mendoza-Silva GM, Belmonte O, Koukofikis A, Nicolau MJ, Costa A, Meneses F, Ebner F, Deinzer F, Vaufreydaz D,
Dao T-K, Castelli E. The smartphone-based offline indoor location competition at IPIN 2016: Analysis and future
work. Sensors. 2017;17(3). https://doi.org/10.3390/s17030557.
18. Potortí F, Park S, Jiménez Ruiz AR, Barsocchi P, Girolami M, Crivello A, Lee SY, Lim JH, Torres-Sospedra J, Seco F,
Montoliu R, Mendoza-Silva GM, Pérez Rubio MDC, Losada-Gutiérrez C, Espinosa F, Macias-Guarasa J. Comparing
the performance of indoor localization systems through the EvAAL framework. Sensors. 2017;17(10). https://doi.org/
10.3390/s17102327.
19. Torres-Sospedra J, Jiménez AR, Moreira A, Lungenstrass T, Lu W-C, Knauth S, Mendoza-Silva GM, Seco F,
Pérez-Navarro A, Nicolau MJ, Costa A, Meneses F, Farina J, Morales JP, Lu W-C, Cheng H-T, Yang S-S, Fang S-H,
Chien Y-R, Tsao Y. Off-line evaluation of mobile-centric indoor positioning systems: The experiences from the 2017
IPIN competition. Sensors. 2018;18(2). https://doi.org/10.3390/s18020487.
20. Goldberg DE. Genetic Algorithms in Search, Optimization and Machine Learning. Reading: Addison-Wesley; 1989.
21. Kennedy J, Eberhart R. Particle swarm optimization. In: Proc. IEEE International Conference on Neural Networks. IV.
Piscataway: IEEE; 1995. p. 1942–8.
22. Kirkpatrick S, Jr. CDG, Vecchi MP. Optimization by simulated annealing. Science. 1983;220(4598):671–80.
23. Finnila AB, Gomez MA, Sebenik C, Stenson C, Doll JD. Quantum annealing: A new method for minimizing
multidimensional functions. Chem Phys Lett. 1994;219(5-6):343–8. https://doi.org/10.1016/0009-2614(94)00117-0.
24. Ali A, Qadir J, Rasool Ru, Sathiaseelan A, Zwitter A, Crowcroft J. Big data for development: applications and
techniques. Big Data Analytics. 2016;1(1):1–24. https://doi.org/10.1186/s41044-016-0002-4.
25. Kingma D, Ba J. ADAM: A method for stochastic optimization. ArXiv e-prints. 2017. arXiv:1412.6980v8 [cs.LG].

Submit your next manuscript to BioMed Central

and we will help you at every step:
• We accept pre-submission inquiries
• Our selector tool helps you to find the most relevant journal
• We provide round the clock customer support
• Convenient online submission
• Thorough peer review
• Inclusion in PubMed and all major indexing services
• Maximum visibility for your research

Submit your manuscript at

www.biomedcentral.com/submit