High-Efficiency Device Localization in 5G Ultra-Dense Networks:

Prospects and Enabling Technologies

Aki Hakkarainen*, Janis Werner*, Mário Costa†, Kari Leppänen† and Mikko Valkama*

*Tampere University of Technology, Finland

†Huawei Technologies Oy (Finland) Co. Ltd, Finland R&D Center

Scope and Outline:

Scope: 5G user node localization, tracking and location prediction Exploiting location information

Outline: •  Properties of 5G ultra-dense networks:

Technical enablers Accurate measurements

•  Fusion of measurements: Collaborative localization

•  Exploiting location-awareness: New location-based services

Properties of 5G Ultra-Dense Networks

Ultra-Dense 5G Networks •  Access nodes (ANs) are

deployed with a very high density à  user nodes (UNs) can see/

hear multiple ANs

[1] METIS, “Deliverable D6.1: Simulation guidelines”, Oct, 2013 [2] Hoydis et al, “Making smart use of excess antennas: massive MIMO,

small cells, and TDD”, Bell Labs Technical Journal, vol. 18, no. 2, Sep, 2013

* Values from Madrid model [1] ** Similar to density in [2]

50 m

Access nodes: density ~ 60m**

Streets: width = 12m*

Housing block: width = 120m*

Figure: LoS probability in METIS’ stochastic channel model [1].

Ultra-Dense 5G Networks

[m]

•  High density à High probability for line of sight (LoS) conditions

[1] METIS, “Deliverable D1.4: METIS channel models”, Feb, 2015

42 m

60 m

[1] P. Kela et al., “A novel radio frame structure for 5G dense outdoor radio access networks”, Proc. IEEE VTC Spring, 2015 [2] T. Levanen et al., “Radio interface evolution towards 5G and enhanced local area communications,” IEEE Access, vol. 2, Sep. 2014. [3] P. Mogensen et al., “Centimeter-wave concept for 5G ultra-dense small cells”, Proc. IEEE VTC Spring, 2014

Proposal [1] (frame duration 163.7 µs):

Proposal [3] (frame duration 250 µs): Proposal [2] (frame duration 500 µs):

5G Small Cell Radio Frames Provide Frequent UL Beacons

•  Use UL beacons/reference symbols for network-centric localization

Properties of 5G Networks à Accurate Measurements •  Ultra-dense network:

High probability for line of sight

•  Radio frames: Frequent uplink pilot signals for uplink channel estimation

•  Bandwidths: Bandwidths are expected to be in the order of 100 MHz and beyond

•  Antenna types: ANs are likely to have antenna arrays or other directional antennas

7

Direction of arrival (DoA)

Time of arrival (ToA)

Time difference of arrival (TDoA)

Received signal strength (RSS)

Example: Direction of Arrival Estimation in a 5G AN

[1] Stoica et al, “The stochastic CRB for array processing: a textbook derivation”, IEEE Signal Process. Lett. vol. 8, no. 5, May 2001

Figure: Cramer Rao bound (CRB) on direction of arrival (DoA) estimation with a circular array and 100 samples. Based on the results in [1].

Fusion of the Measurements

Measurements from ANs are fused into a location estimate

10

5G Ultra-Dense Networks à High Localization Accuracy •  Fusion of various accurate measurements across multiple

measurement/observation points

à extremely high localization accuracy is expected (even in sub-meter range [1-2])

•  Compare to the accuracies of the existing localization systems –  OTDoA* in LTE: a few tens of meters [3] –  GNSS: around 5 meters [4] –  WLAN fingerprinting: 3-4 meters [5]

*OTDoA = Observed time difference of arrival

[1] 5G Forum, “5G white paper: New wave towards future societies in the 2020s,” Mar. 2015. Available: http://www.5gforum.org/ [2] NGMN Alliance, “5G White Paper,” Feb. 2015. Available: https://www.ngmn.org/uploads/media/NGMN 5G White Paper V1 0.pdf [3] J. Medbo, I. Siomina, A. Kangas, and J. Furuskog, “Propagation channel impact on LTE positioning accuracy: A study based on real

measurements of observed time difference of arrival,” in Proc. IEEE PIMRC, Sep. 2009 [4] D. Dardari, P. Closas, and P. M. Djuric, “Indoor tracking: Theory, methods, and technologies,” accepted for publication in IEEE Trans.

Vehicular Tech., 2015. [5] H. Liu et al., “Push the limit of WiFi based localization for smartphones,” in Proc. 18th Annu. Int. Conf. Mobile Computing and

Networking (MobiCom), 2012

Example: Fusion of DoA Estimates from 4 ANs

Figure: Localization CRB on the localization RMSE for example geometry. 4 ANs, each equipped with 10 antennas.

RMSE �p

CRBx

+CRBy

=D�

'p2

Figure: Example geometry.

12

Tracking and Location Prediction

13

•  Accurate, fast and “always on” UN localization enables UN tracking

•  Tracking, combined with predictive estimation algorithms, such as extended Kalman filters, enables UN location prediction

Predicted trajectory

Estimated trajectory

Current location

Estimated velocity

Time for a short video clip…

•  Video available in: http://www.tut.fi/5G/VTC15

Prospects of 5G Localization

Prospects of 5G Localization

16

•  Localization can be carried out on the network side using UN/uplink reference signals à  “always on” network-centric localization without

heavy load on UN batteries

•  Tracking and prediction of UN positions à  environment learning

•  New markets for network operators: self driving cars, VTX, robots, intelligent traffic systems, …

•  Advanced radio network functionalities: e.g. proactive radio resource management and mobility management

Prospects #1 Radio Environment Maps

17

•  Radio environment maps can be used, e.g., for proactive

radio resource management

REM processing

location estimates +

measured radio parameters

Prospects #2 Prediction and Cognitive Localization

•  Human mobility depends highly on historical behavior and is predictable up to 88 % [1] à mobility database

[1] X. Lu, et. Al. “Approaching the limit of predictability in human mobility,” Nature: Scientific Reports, vol. 3, no. 2923, Oct. 2013.

Mobility database (inc. e.g. most common routes,

traffic patterns)

Location information (inc. current location, tracking)

Machine learning type of algorithms

UN localization & movement prediction with a higher accuracy and for a longer period of time

•  Assume that we are tracking a car and can predict it’s movement •  Then it is possible to allocate radio resources proactively

Prospects #3 Predictive Radio Resource Management

19

Point A à AN 1 Point B à AN 2

Prospects #4 Content Prefetching

20

•  Video streaming in a car with a predicted route •  ANs 3 and 4 are congested because of a match in a stadium

à this would result in a poor quality of service (QoS) •  However, the user can be prioritized while still under ANs 1 and 2 with free

resources à proactive content delivery to a buffer à good user experience

Prospects #5 Routing in the Backhaul

21

•  Network observes a truck which would block the visibility from AN 1 to the user in a car à proactive re-routing of the user data to AN 2 à seamless service and better user experience

•  Properties of 5G ultra-dense networks enable –  network-centric UN localization with a very high accuracy –  frequently updated UN tracking –  UN location prediction

•  This location-awareness can be utilized in the UN, ANs or by third parties to provide new location-based services such as –  Radio environment maps –  Cognitive localization and prediction –  Proactive radio resource management –  Content prefetching –  Routing in the backhaul

Conclusions

22

Thank You!

Aki Hakkarainen*, Janis Werner*, Mário Costa†, Kari Leppänen† and Mikko Valkama*

*Tampere University of Technology, Finland

†Huawei Technologies Oy (Finland) Co. Ltd, Finland R&D Center