Localization System based on Ultra Low-power Radio Landmarks

Fabian Höﬂinger

, Joan Bordoy

, Rui Zhang

, Amir Bannoura

, Nikolas Simon

, Leonhard Reindl

and Christian Schindelhauer

Department of Microsystems Engineering, University of Freiburg, Germany

Department of Computer Science, University of Freiburg, Germany

Keywords:

Indoor Localization, Ultra-wideband, On-demand, Wake-up, Low Power.

Abstract:

In this paper we present a novel indoor localization system using external reference landmarks as a guidance

system for emergency responders. The landmarks are based on low-power wake-up nodes which can be

integrated into smoke detectors. The radio wake-up technology is equipped in the system to extend the lifetime

of landmarks. While in sleep mode our landmarks have an overall power consumption of 66 µW making them

ready-to-use in case of an emergency for up to 5 years. The landmarks are small and cost-efﬁcient and may

be integrated into the building infrastructure. The positioning is achieved by combining the radio ranging and

IMU based dead reckoning to overcome the disadvantages of both systems. The experimental results show that

the proposed system is able to outperform both standalone systems and meanwhile maintain the low power

consumption.

1 INTRODUCTION

In the recent decade, a growing demand in precise

indoor locating systems could be observed (Bordoy

et al., 2016)-(Kuhn et al., 2009) wireless so that in-

door location services, such as locating victims in

avalanches or earthquakes, injured skier on ski slope,

military personnel, ﬁre ﬁghters or lost children, can be

delivered. However, in contrast to this increasing de-

mand, the technology for reliable indoor navigation is

still in its infancy, since these applications need very

high accuracy requirements, low power consumption

and low complexity. Nowadays most of indoor locat-

ing technologies can be divided into acoustic, optical,

and radio frequency methods. The last type of meth-

ods can be divided into continuous wave, for example,

WLAN or RFID, and impulse signals. Unfortunately,

the above mentioned technologies either cannot fulﬁll

the criteria of high accuracy or low power consump-

tion required by indoor location service applications.

2 STATE-OF-THE-ART

Many non-GPS localization systems based on vari-

ous technologies have been developed (Fischer and

Gellersen, 2010). Most of them can be classiﬁed into

absolute and relative localization systems.

Absolute localization systems normally require

external references that consists of ﬁxed landmarks

such as Wi-Fi access points (Bahl and Padmanabhan,

2000) or ultra-wide band systems (Kuhn et al., 2009)

to determine the position by measuring the Received

Signal Strength Indicator (RSSI) or the Time of Ar-

rival (ToA)/Difference of Arrival (TDoA). Due to its

high energy consumption, such systems are required

either to be connected to the power grid or frequent

battery charging/replacement. As a result, such sys-

tem are not suitable for catastrophic scenarios due to

its high installation costs and power consumption.

The most commonly used relative non-GPS in-

door localization approach is inertial measurement

units (IMU) based dead reckoning. The IMU can be

attached to the body or mount on the shoe of the res-

cue forces (Höﬂinger et al., 2013; Zhang et al., 2013;

Höﬂinger et al., 2012; Nilsson et al., 2014). In this

approach, the relative positioning is obtained in a re-

cursive manner, i.e. the direction and the distance rel-

ative to the initial state are calculated via integration

of acceleration and gyroscope data. Therefore, no

external reference or pre-installation is needed. The

system can also be powered by small size batteries.

The main drawback of such systems is that the error

will be accumulated over time due to drift of the sen-

sors. Therefore, several approaches have been devel-

oped (Zhang et al., 2011), (Fang et al., 2005), (Zhang

and Reindl, 2011) to minimize such error. Neverthe-

less, standalone IMU based localization systems are

not capable of providing sufﬁcient accuracy for long

term measurements, especially if the nature of move-

Höﬂinger, F., Bordoy, J., Zhang, R., Bannoura, A., Simon, N., Reindl, L. and Schindelhauer, C.

Localization System based on Ultra Low-power Radio Landmarks.

DOI: 10.5220/0006608800510059

In Proceedings of the 7th International Conference on Sensor Networks (SENSORNETS 2018), pages 51-59

ISBN: 978-989-758-284-4

ments is unsteady, which is often the case during res-

cue operations.

In order to fulﬁll the requirements of the indoor lo-

cation application such as very high accuracy require-

ments, low power consumption and low complexity,

one should decrease the system energy consumption

especially for absolute localization systems and in-

crease the tracking accuracy of the system. By ap-

plying wake-up technology, the power consumption

can be signiﬁcantly reduced. By combining both ab-

solute and relative localization systems, the tracking

accuracy can be greatly enhanced.

3 CONCEPT OVERVIEW

In this paper we present a indoor localization system

using landmarks based on low-power wake-up nodes

which can be integrated into smoke detectors. As a

central component of this system we have developed a

handheld device that serves as a master node to com-

municate with our landmarks. The handheld device

broadcasts a wake-up message and measures the Re-

ceived Signal Strength Indicator (RSSI) of each an-

swer received from all landmarks that woke up. Based

on this data the master node calculate the current po-

sition of the handheld device.

Furthermore, the handheld device is able to re-

ceive inertial sensor data of our wireless IMU which

can be integrated in shoes. The additional informa-

tion allows movement tracking between two wake-up

events to increase localization accuracy. Moreover,

due to the high short-distance accuracy of inertial data

based localization, the number of wake-up events can

be reduced and hence lifetime of the reference land-

marks is increased.

4 HARDWARE DESIGN

In the following section the main components of the

hardware are described:

4.1 Handheld Device

The developed prototype of our handheld device

consists of two parts: The credit-card sized low-

power computer BeagleBone Black with a compati-

ble touchscreen and our developed expansion circuit

board.

As shown in Fig. 1 our developed expansion cir-

cuit board. is made up of two wireless modules, a

power management module and an EEPROM. One of

Figure 1: Block diagram of the expansion board with its

components. The board is stacked on the BeagleBone Black

for communication and power supply (Simon et al., 2015).

Figure 2: Expansion board stacked on top of the credit-card

sized low-power computer BeagleBone Black (Simon et al.,

2015).

the wireless modules communicates with our wake-

up nodes by transmitting wake-up messages (wake-

up message details see Gamm et al. (Gamm et al.,

2012)) if requested by the computer and receiving the

answers of the landmarks. The other one receives in-

ertial sensor data from our wireless IMU. Both wire-

less modules use a CC430 low-power microcontroller

from Texas Instruments to communicate on a fre-

quency of 868 MHz with the appropriate component.

To extend the wake-up range of the system an ad-

ditional front-end ampliﬁer CC1190 is used for the

wireless module. Furthermore, each controller uses

a separate UART connection to transfer the received

data to the BeagleBone Black computer via the cape

expansion connectors.

4.2 Low-power Wake-up Landmarks

Our developed landmarks are based on a wake-up

technology presented in Gamm et al. (Gamm et al.,

2012) which uses a 125 kHz wake-up receiver. Low

power wake-up receivers are used for keyless go entry

SENSORNETS 2018 - 7th International Conference on Sensor Networks

Figure 3: Topview of our developed landmark integrated

into a commercially available smoke detector.

Impedance

Matching

Rectifier Lowpass

AS3932

Wake-Up

Receiver

MSP430

Wake-Up

CC1101

868 MHz

Transceiver

Antenna

Switch

Control

Figure 4: Block diagram of the wake-up circuit in our land-

marks (Gamm et al., 2012).

systems in automotives. They are build to work a long

time without a battery change and therefore operate at

low frequencies. The short wake-up range of about 3

meters due to inductive coupling is of no limiting fac-

tor for the keyless go application.

In Fig. 4 a block diagram of the wake-up landmark

is shown. When the node is in active mode, the an-

tenna switch is conﬁgured so that all in and outgo-

ing signals pass to the main radio transceiver chip.

Before entering the sleep mode, the microcontroller

toggles the antenna switch. All incoming signals dur-

ing sleep mode are then routed to the analog circuit

consisting of impedance matching, rectifying and low

pass ﬁltering. The incoming 868 MHz wake-up signal

is passively demodulated by a rectiﬁer and ﬁltered.

The passiv demodulation and the analog path is an

important factor in the performance of the wake-up

receiver since non-ideal impedance matching will re-

sult in a shorter wake-up distance. The analog path

of the presented node consists of a matching network,

two demodulation diodes and a low pass ﬁltering cir-

cuit. The RF Schottky demodulator diodes are con-

nected as a typical voltage doubler circuit. Its pur-

pose is to rectify the modulated RF carrier signal. Be-

cause of the OOK modulation of the carrier signal

the rectiﬁer charges a capacitor of a low pass ﬁlter

up to a certain value during the ON period of the car-

rier. When the carrier is turned OFF the capacitor is

discharged through a resistor. This way, a triangu-

lar signal is generated with a frequency of 125 kHz.

Afterwards, the signal is coupled to the wake-up re-

ceiver through a capacitor in order to remove any DC

offset. The ﬁltered signal is then passed to the input

of the wake-up receiver IC. In our node we used the

AS3932 wake-up receiver from Austriamicrosystems.

It consumes in one channel listening mode 2.7 µA cur-

rent and has a wake-up sensitivity of 100 µVRMS as

well as a high input impedance of 2 MΩ. One of the

main reasons for choosing the AS3932 is that it has

an integrated correlator which compares the received

signal to a byte pattern saved in a conﬁguration reg-

ister. In case of a positive correlation of the incom-

ing signal with an internal saved 16-Bit sequence the

wake-up receiver changes the state of one of its output

pins. This signal change is fed to an interrupt capable

input port of the microcontroller. The generated in-

terrupt triggers the controller from its sleep to active

mode. When entering the active mode, the controller

again toggles the antenna switch so that the main ra-

dio transceiver is connected to the antenna. The node

can then establish a normal communication link, e.g.

send an acknowledge or send a message for the RSSI-

Measurement between landmark and the handheld de-

vice.

While listening for a wake-up packet the standby

current of the node is about 2.78 µA which results

in a standby power consumption of 5.6 µW. Using

a CR2032 coin cell battery with a capacity Q

Bat

230 mAh as power supply we have to take an addi-

tional self-discharge current of about 263 nA into ac-

count. Therefore, the overall current consumption of

the node in sleep mode sums up to 3.044 µA. After

the node has been in this mode for t

sleep

the remain-

ing charge of the battery Q

Le ft

can be calculated with

the following equation 1. Therefore, the theoretical

maximum lifetime of the node without any wake-up

is 8.62 years.

Le ft

= Q

Bat

−t

sleep

· I

sleep

(1)

Assuming a maximum current consumption of

15 mA during a sending process which takes about

13 ms the theoretical maximum operating time of our

landmarks after t

sleep

can be calculated using equation

maxOp

Le ft

· T

Wakeup

Send

· I

Send

+ (T

Wakeup

− T

Send

) · I

Sleep

(2)

Once the system is in operation the wake-up pe-

riod T

Wakeup

dominates the power consumption and

thus the maximum life time of our nodes (Gamm

et al., 2012). Figure 5 shows the negative linear be-

Localization System based on Ultra Low-power Radio Landmarks

Low-power sleeping time T

sleep

(years)

0 1 2 3 4 5 6 7 8 9

Maximum operating time T

(days)

100

0.5 Hz Wake-up

1 Hz Wake-up

2 Hz Wake-up

Figure 5: This ﬁgure shows the maximum theoretical oper-

ating time of a landmark for different wake-up frequencies

after it has been in its ultra-low power state for T

sleep

Table 1: Temperature range of critical components.

Operating temperature

Component Min. (

◦

C) Max. (

◦

MSP430F2350 -40 85

CC1101 -40 85

ADG918 -40 85

AS3932 -40 85

HSMS285C -65 150

Balun 868 -40 125

Crystal -10 70

haviour of the remaining operating time for three dif-

ferent wake-up periods after the node has been sleep-

ing for t

sleep

As a guidance system for emergency responders

the operating temperature range is an important crite-

ria. Table 1 shows that our crystal is the most critical

component which limits operation theoretically to a

temperature range of −10

◦

C to 70

◦

C. However, our

practical tests have shown that a successful communi-

cation with our nodes is possible within a temperature

range of −20

◦

C to 115

◦

C as shown in ﬁgure 6. No-

tice that the radio frequency from the quartz oscillator

changes when the temperature varies. Beyond 100

◦

the reception bandwidth of receiver can not detect the

transmitted radio frequency any more.

With this technology we build a real-time capa-

ble low-power landmark with a theoretical maximum

standby time which is comparable with the one of

commercially available smoke detectors. Therefore,

we adapted our circuit board design to be able to in-

tegrate the nodes in this extisting infrastructure and

hence ease the hardware setup to have a system which

is ready-to-use in case of a catastrophic scenario (ﬁg-

ure 3).

The sensitivity has been measured in by using a

signal generator. A successful wake-up was observed

up to an attenuation of -52 dBm (Gamm et al., 2012).

Through improved impedance matching a wake-up

Temperature (°C)

-20 0 20 40 60 80 100 120

Successful communication (%)

100

Figure 6: Our measurements show a successful communi-

cation within a temperature range of −20

◦

C to 115

◦

distance of up to 80 meters is possible at 20 dBm

power output (Gamm et al., 2013).

4.3 Micro-inertial Measurement Unit

(IMU)

Our wireless micro-IMU V3 (Fehrenbach, 2014) used

in this application, has already been successfully used

for short-distance indoor motion tracking of pedestri-

ans when mounted on a shoe (Höﬂinger et al., 2012).

With its small size of 22 mm×14 mm× 4 mm the mi-

cro IMU is in this application mounted on a shoe

and transmits its sampled sensor data wirelessly to

our receiver, the handheld device. Concerning this,

a CC430 microcontroller from Texas Instruments is

used to transmit the data at 868 MHz. Besides the

controller the micro IMU consists of a three-axis ac-

celerometer, a three-axis gyroscope and a three-axis

magnetometer as well as a voltage regulator (see Fig.

7). The raw data of the sensors can be sent with a

maximum rate of 640 samples per second. Thereby,

data post processing is done by the receiver to in-

crease the performance of the IMU. More details

about metrological characteristics can be found in

Höﬂinger et al. (Höﬂinger et al., 2012).

5 LOCALIZATION

5.1 Problem Setting

The low power wake-up nodes are placed randomly

at unknown stationary positions S

(1 ≤ j ≤ B). For

simplicity, we assume they are located in a two-

dimensional Euclidean space. The handheld device H

moves in the two-dimensional Euclidean space, wak-

ing up the nodes and measuring the signal strength

(RSSI).

SENSORNETS 2018 - 7th International Conference on Sensor Networks

Antenna

868MHz

CC430

Microcontroller

RF- transceiver

Accelerometer

MMA8451Q

Gyroscope

ITG3200

Magnetometer

HMC6343

Voltage

Regulator

I²C

4- 6.5 V

Figure 7: Block diagram of the Micro-IMU V3 (Fehren-

bach, 2014). The IMU is capable of transmitting acceler-

ation, magnetic ﬁeld and angle velocity sensor data via its

868 MHz radio module.

Figure 8: Topview of the micro-IMU V3 which is mounted

on a shoe.

5.2 Range Estimation

The handheld device is located at a distance d from

the node j:

d = kH − S

k (3)

where k · k denotes the Euclidean norm.

The relation between d and the RSSI measure-

ments can be modelled as follows (Qi, 2003):

4π

(4)

where P

is the transmitted power, G

and G

are the

transmitter and the receiver gains, respectively, n is

the path loss exponent, and g and γ are the parame-

ters that conform the Rayleigh/Rician and lognormal

distributions, respectively.

Assuming the received signals are averaged over

a certain time interval, the fast fading term can be

eliminated. Thus, the logarithmic equation which re-

lates the received signal strength and the distance can

be formulated as follows (Qi, 2003) (Mazuelas et al.,

2009):

(dBm) = α − 10nlog

(d) + χ (5)

Figure 9: Schematic of the under-determined equation sys-

tem. If the ﬁreﬁghter moves continuously, for every new

measurement there are two new variables to estimate only

for its position. On the other hand, if he stops, his device

receives one signal from every node (S

and S

), leading

to three constrains for every two position variables (Simon

et al., 2015).

where χ denotes a Gaussian random variable with

zero mean caused by shadowing. The term α is a

constant which depends on the averaged slow and fast

fading, the transmitted power and the gains of the an-

tennas.

5.3 Node and Standing Positions

Localization

The continuous movement of the master device re-

sults in a system of equations which cannot be

solved in closed form, as for every received measure-

ment there are two new variables of position to es-

timate. Consequently, the equation system is under-

determined and cannot be solved in closed form with-

out further information or assumptions on the sce-

nario. Therefore, we assume the master node stops

in q different positions H

, then we have time to re-

ceive at least one signal from every node (stop-and-

go motion). Doing this, it is only required to estimate

one handheld device position (2 variables) for every

B received signals, which makes possible an uniquely

determined system of equations (cf. Fig. 9).

Then, we obtain a system of hyperbolic equations

of the form:

p, j

= kH

− S

k − z

p, j

(6)

where 1 ≤ j ≤ B and 1 ≤ p ≤ q. The term z

p, j

the measured distance between the sender j and the

standing position p using a RSSI measurement and

Equation (5).

The system of equations has now qB independent

equations, which has to be higher than the number of

Localization System based on Ultra Low-power Radio Landmarks

1 2 3 4 5 6 7 8 9

Distance (m)

Signal Strength (dBm)

Node 1

Node 2

Node 3

Node 4

Node 5

Logarithmic Equation

Figure 10: Real and modeled relation between the received

signal strength and the distance between the node and the

hand-held device. The uncertainty bars show the standard

deviation of the signal strength after the 5% highest and

the 5% lowest RSSI signals for each distance have been re-

jected.

variables:

qB ≥ 2q

|{z}

Handheld device

+ 2B

|{z}

Nodes

(7)

Which means the system of equations can be solved in

a closed form if the number of standing still positions

q is higher than:

q ≥

B − 2

(8)

Assuming the stop-and-go motion and having a

number of standing positions and nodes fulﬁlling

Equation (8) the system of hyperbolic equations can

be solved with local optimization algorithms. We

use both the gradient descent and the Gauss-Newton

method, the two are ﬁrst-order methods that use the

derivative of the system of hyperbolic error equa-

tions. The Equation (6) results in a quadratic objective

which can be formulated as follows:

∑

p=1

∑

j=1

arg min

( f

p, j

)

. (9)

Which in vector notation is proportional to w =

with b = ( f

1,1

, ..., f

q,B

)

. The operator (·)

denotes

the transposition.

We calculate the direction of the steepest ascent:

Ow = O





= Q

b (10)

where Q is the Jacobian matrix:

Q =







∂ f

1,1

∂S

. . .

∂ f

q,B

∂S

∂ f

1,1

∂S

. . .

∂ f

q,B

∂S

∂ f

1,1

∂H

. . .

∂ f

q,B

∂H

∂ f

1,1

∂H

. . .

∂ f

q,B

∂H







(11)

The partial derivative with respect to a vector is

deﬁned as the derivative with respect to each of its

components:

∂ f

p, j

∂H



∂ f

p, j

∂H

p,x

∂ f

p, j

∂H

p,y



(12)

In our case the partial derivative with respect to

the node position S

is:

∂ f

p, j

∂S

= −

− S

(13)

The partial derivative with respect to the handheld po-

sition is:

∂ f

p, j

∂H

− S

(14)

All the variables which need to be estimated are

components of the state vector u:

u = (S

, ..., S

, H

, ..., H

)

(15)

Every iteration the state vector is updated using Q

and b. The methods used are:

5.3.1 The Gradient Descent Method

In every iteration step l the Gradient Descent method

updates the state vector in direction of the steepest

descent. The adaptive factor λ sets the step width.

u = λOw = λQ

l+1

= u

−

(16)

5.3.2 The Gauss-Newton Algorithm

Instead of relying on an adaptive factor γ it calculates

the step size using the inverse (Q

−1

for every it-

eration:

u = (Q

−1

b) (17)

We calculate for higher numerical stability the

pseudo-inverse with singular value decomposition in-

stead of calculating the inverse.

This algorithm is faster, nevertheless it is very

prone to divergence when applied to random initial

positions. However, it can be used when the Gradient

Descent error function has become steady to reduce

notably the number of iterations (Wendeberg et al.,

2013).

5.4 Handheld Device Localization. Data

Fusion

The IMU has been proved to be capable of tracking

pedestrians in indoor areas showing a maximum de-

viation of 1 m after a walk of 30 m (Höﬂinger et al.,

SENSORNETS 2018 - 7th International Conference on Sensor Networks

2012). However, it cannot be used as the only source

of information due to its accumulative error. In or-

der to solve this, we combine the measurements of

the IMU and the anchor nodes using an unscented

Kalman ﬁlter (UKF). The UKF is a recursive state es-

timator which fulﬁls the bayesian ﬁltering model and

uses a set of sample points (sigma points) to linearise

non-linear functions. Therefore, it is cheaper in com-

putation than other similar algorithms like the parti-

cle ﬁlter, which requires evaluation of a large number

of particles, or the extended Kalman ﬁlter, which re-

quires calculation of the Jacobian matrix. More de-

tailed information about it and its implementation can

be found in (Thrun et al., 2005). In our case, the state

vector x

which contains the variables to estimate has

the following components:



, V

, A



(18)

where M

is the position of the target, V

his veloc-

ity and A

the acceleration. All of them in a two-

dimensional euclidian space.

We use the the Weiner process acceleration model

(Bar-Shalom et al., 2001) in two dimensions.









= Θ

t−1





t−1





+Φ

t−1

∼ N (0, R

)

(19)

where





1 ∆t ∆t

0 1 ∆t

0 0 1





= τ







∆t







(20)

where τ is a parameter that depends on the expected

movement of the target.

In this case we assume each time the nodes are

woken up the RSSI measurements are received at the

same position. Then, having RSSI measurements of

N different nodes at time t, the ﬁrst N components

of the predicted measurement vector z

t,1:N

fulﬁl the

following sensor model:

t,i

= α−10nlog

(kM

−S

k)+ρ

∼ N (0, σ

)

(21)

where σ

is the expected standard deviation of the

RSSI measurement noise. We combine these mea-

surements with the foot-mounted IMU measure-

ments. The sensors that we use are the accelerometer

and the gyroscope.

We remove the effect of the gravity and extract

the x and y components of the acceleration by com-

bining the acceleration and the angular rate. More

H7H6H5H4H3H2H1

Start

End

Correct landmark position

Estimated landmark position

Standing position

Figure 11: Node localization. The master node stands still

in 7 positions H

,...,H

measuring the signal strength of the

nodes S

,...,S

. These measurements are used as an input of

local optimization algorithms (Simon et al., 2015).

information about this transformation can be found in

(Kuipers, 2002). Then, the components N + 1 and

N + 2 of the measurement vector are predicted as fol-

lows:

t,N+1:N+2

= A

+ ν

∼ N (0, σ

) (22)

where σ

is the expected noise of the acceleration

measurement.

To reduce the drift of the IMU sensors, we detect

when the human being is not moving and we set the

velocity and the acceleration to zero, as it is done in

(Woodman and Harle, 2008). As the IMU has a much

higher sampling rate than the nodes, the sensor data

fusion is only done when the velocity and acceleration

are not set to zero and there is a RSSI measurement

available. In the other cases, the UKF estimates the

values using only the IMU measurements.

6 EXPERIMENTAL RESULTS

6.1 Relation between Signal Strength

and Distance

The parameters of the theoretical model which re-

lates the received signal strength and the distance (see

Equation (5)) are estimated by collecting measure-

ments from 5 nodes in 7 different positions in a corri-

dor. The best ﬁt to the real measurements are a path

loss exponent of 7.1 and a constant α equal to 89.7

(see Fig. 10).

6.2 Node and Standing Positions

Localization

In order to test the node localization, we stop in 7 po-

sitions H

, ..., H

, as shown in Fig. 11. Such positions

are assumed to be known, as they can be estimated by

Localization System based on Ultra Low-power Radio Landmarks

-

)



Node 1

Node 2

Node 3

Node 4

Node 5

1 m

Start/End

Data fusion

IMU

Real trajectory

Figure 12: Master node localization. A person moves con-

tinuously with the IMU attached to his shoe. Sensor data

fusion is performed to combine the RSSI and the IMU data.

the inertial measurement unit, when the accumulative

error is still negligible. As the standing positions are

symmetrical to the node positions, the local optimiza-

tion algorithms are initialized on the side of their true

value. The resulting median error is 78 cm.

The same data is used to test the hand-held device

localization without IMU. The node positions are as-

sumed to be known and the target is located with a

median error of 99 cm.

6.3 Continuous Movement Tracking

In order to test the performance of the sensor data fu-

sion we perform an experiment in a different building

of the faculty, using the same path loss exponent and

α mentioned above. The device was attached to the

foot of a person. The sampling rate of the IMU is

50 Hz and data from the wake-up nodes is received

every 3 seconds. As having only one RSSI measure-

ment can lead to high errors, the measurement noise

of the RSSI measurements is increased in this case in

order to reﬂect this uncertainty. In Fig. 12 we can

see both the result of using only the IMU and sensor

data fusion. The median error using only the IMU

is 0.470 m with a standard deviation of 0.332 m while

the median error using also the RSSI measurements is

0.276 m with a standard deviation of 0.229 m. There-

fore, the error is notably reduced.

7 CONCLUSION AND FUTURE

WORK

In this paper, we have presented a novel self-

calibrating indoor localization system for emergency

responders using 868 MHz radio landmarks and iner-

tial sensor data. For this system we have developed

new wireless landmarks using ultra low-power wake-

up technology, which makes them ready-to-use for up

to 8 years if powered by a coin cell. The nodes are

integrable into building infrastructures like smoke de-

tectors. Moreover, a handheld device has been devel-

oped to send initial wake-up calls to the landmarks,

measure the RSSI of the response, and use this data to

estimate and display the current position of the ﬁre-

ﬁghter. Additionally, our handheld device is able to

receive inertial sensor data by a body-mounted micro-

intertial measurement unit (IMU) to increase localiza-

tion accuracy. The data is fused with an Unscented

Kalman ﬁlter.

The experimental results demonstrate that using

the obtained relation of signal strength and distance

the system is able to locate our landmarks with a me-

dian error of 78 cm in an indoor environment. More-

over, fusing the RSSI and IMU data, the continous

trajectory of the ﬁreﬁghter is tracked with a median

error of 27.6 cm.

Due to the fact that the RSSI approach is highly

sensitive to the environments, long-term high accu-

racy tracking can not be ensured. In the future,

we plan to equip our landmark device with Ultra-

Wide-Band (UWB) technology. Instead of measur-

ing the RSS, signal time of arrival will be used as

the measurement for improving the localization accu-

racy. Meanwhile, more reliable estimators that explic-

itly consider outlier error mitigation, e.g. RANSAC

(Bordoy et al., 2016), or robust regression (Bordoy

et al., 2017) will be investigated and adapted for the

proposed system so that system robustness can be en-

hanced.

ACKNOWLEDGEMENTS

This work has partly been supported by the German

Federal State Postgraduate Scholarships Act (Landes-

graduiertenförderungsgesetz - LGFG) within the co-

operative graduate school “Decentralized sustainable

energy systems".

SENSORNETS 2018 - 7th International Conference on Sensor Networks

REFERENCES

Bahl, P. and Padmanabhan, V. N. (2000). Radar: An in-

building rf-based user location and tracking system.

In INFOCOM 2000. Nineteenth Annual Joint Confer-

ence of the IEEE Computer and Communications So-

cieties. Proceedings. IEEE, volume 2, pages 775–784.

Ieee.

Bar-Shalom, Y., Li, X. R., and Kirubarajan, T. (2001). Es-

timation with applications to tracking and navigation.

Wiley-Interscience.

Bordoy, J., Schindelhauer, C., Zhang, R., Höﬂinger, F., and

Reindl, L. M. (2017). Robust extended kalman ﬁlter

for nlos mitigation and sensor data fusion. In Inertial

Sensors and Systems (INERTIAL), 2017 IEEE Inter-

national Symposium on, pages 117–120. IEEE.

Bordoy, J., Wendeberg, J., Schindelhauer, C., Höfﬂinger,

F., and Reindl, L. M. (2016). Exploiting ground re-

ﬂection for robust 3d smartphone localization. In In-

door Positioning and Indoor Navigation (IPIN), 2016

International Conference on, pages 1–6. IEEE.

Fang, L., Antsaklis, P. J., Montestruque, L., McMickell,

M. B., Lemmon, M., Sun, Y., Fang, H., Koutroulis,

I., Haenggi, M., Xie, M., et al. (2005). Design of a

wireless assisted pedestrian dead reckoning system-

the navmote experience. Instrumentation and Mea-

surement, IEEE Transactions on, 54(6):2342–2358.

Fehrenbach, P. (2014). Entwicklung einer inertialsensorik

zur analyse von schwimmbewegungen.

Fischer, C. and Gellersen, H. (2010). Location and navi-

gation support for emergency responders: A survey.

IEEE Pervasive Computing, 9(1):38–47.

Gamm, G., Kostic, M., Sippel, M., and Reindl, L. M.

(2012). Low power sensor node with addressable

wake-up on demand capability. Int. J. Sensor Net-

works, 11(1):48–56.

Gamm, G., Sester, S., Sippel, M., and Reindl, L. M. (2013).

Smartgate - connecting wireless sensor nodes to the

internet. Journal of Sensors and Sensor Systems,

pages 45–50.

Höﬂinger, F., Müller, J., Zhang, R., Reindl, L. M., and Bur-

gard, W. (2013). A wireless micro inertial measure-

ment unit (imu). Instrumentation and Measurement,

IEEE Transactions on, 62(9):2583–2595.

Höﬂinger, F., Zhang, R., and Reindl, L. M. (2012). Indoor-

localization system using a micro-inertial measure-

ment unit (imu). In European Frequency and Time

Forum (EFTF), 2012, pages 443–447. IEEE.

Kuhn, M., Zhang, C., Lin, S., Mahfouz, M., and Fathy, A. E.

(2009). A system level design approach to uwb lo-

calization. In Microwave Symposium Digest, 2009.

MTT’09. IEEE MTT-S International, pages 1409–

1412. IEEE.

Kuipers, J. (2002). Quaternions and Rotation Sequences .

Princeton Paperbacks.

Mazuelas, S., Bahillo, A., Lorenzo, R. M., Fernandez, P.,

Lago, F. A., Garcia, E., Blas, J., and Abril, E. J.

(2009). Robust indoor positioning provided by real-

time RSSI values in unmodiﬁed wlan networks. Se-

lected Topics in Signal Processing, IEEE Journal of,

3(5):821–831.

Nilsson, J.-O., Rantakokko, J., Händel, P., Skog, I., Ohls-

son, M., and Hari, K. (2014). Accurate indoor po-

sitioning of ﬁreﬁghters using dual foot-mounted iner-

tial sensors and inter-agent ranging. In Proceedings

of the Position, Location and Navigation Symposium

(PLANS), 2014 IEEE/ION.

Qi, Y. (2003). Wireless geolocation in a non-line-of-sight

environment. PhD thesis, Princeton University.

Simon, N., Bordoy, J., Höﬂinger, F., Wendeberg, J., Schink,

M., Tannhäuser, R., Reindl, L. M., and Schindel-

hauer, C. (2015). Indoor localization system for emer-

gency responders with ultra low-power radio land-

marks. In International Instrumentation and Measure-

ment Technology Conference (I2MTC).

Thrun, S., Burgard, W., and Fox, D. (2005). Probabilistic

Robotics. mitpress.

Wendeberg, J., Höﬂinger, F., Schindelhauer, C., and Reindl,

L. (2013). Calibration-free TDOA self-localization.

Journal of Location Based Services.

Woodman, O. and Harle, R. (2008). Pedestrian localisa-

tion for indoor environments. In Proceedings of the

10th international conference on Ubiquitous comput-

ing, pages 114–123. ACM.

Zhang, R., Höﬂinger, F., Gorgis, O., and Reindl, L. (2011).

Indoor localization using inertial sensors and ultra-

sonic rangeﬁnder. In Wireless Communications and

Signal Processing (WCSP), 2011 International Con-

ference on, pages 1–5. IEEE.

Zhang, R., Höﬂinger, F., and Reindl, L. (2013). Inertial sen-

sor based indoor localization and monitoring system

for emergency responders. Sensors Journal, IEEE,

13(2):838–848.

Zhang, R. and Reindl, L. (2011). Pedestrian motion based

inertial sensor fusion by a modiﬁed complementary

separate-bias Kalman ﬁlter. In Sensors Applications

Symposium (SAS), 2011 IEEE, pages 209–213. IEEE.

Localization System based on Ultra Low-power Radio Landmarks