Shared Networks and the Evolu;on towards 5G

Transcription

1 Shared Networks and the Evolu;on towards 5G Luiz DaSilva Professor of Telecommunica:ons, Trinity College ICNC 2018 Maui, HI, 5-8 March 2018

2 Trinity College Dublin

3 CONNECT Future Communica:ons and Networks X Rapid Prototyping and Experimenta;on Media Rich Applica;ons M2M/D2D applica;ons Audio-visual media processing service plajorms privacy/security services cloud services mobile services network performance monitoring network op;miza;on virtualiza;on techniques cogni;ve networking op;cal/wireless interface op;cal architectures cyberphysical systems sensor networks wireless/mobile architectures spectrum management sohware/cogni;ve radio plajorms PHY layer signal processing PHY layer monitoring RF design antennas op;cal technologies thermal strategies energy harves;ng strategies microelectronic circuits smart sensors

4 5G requirements Enhanced Mobile Broadband DENSE NETWORKS MOVING NETWORKS SHARED NETWORKS 5G Massive Machine Communica;ons LOW ENERGY NETWORKS Ultra reliable, low latency CONVERGED NETWORKS NANO NETWORKS

5 How do we get there? Network slicing Virtualisa:on Sharing C. Sexton, N. Kaminski, J. Marquez Barja, N. MacheU, and L. A. DaSilva, 5G:Adaptable Networks Enabled by Versa:le Radio Access Technologies, IEEE Communica;ons Surveys and Tutorials, 2017.

6 Greater versa;lity and flexibility in Duplexing, waveforms, propaga;on in new bands, massive MIMO Spectrum, RAN, processing, storage, energy consump;on Spectrum policy, net neutrality, privacy, compe;;on Service models, provider plans, sharing agreements, virtual and micro operators

7 Spectrum and infrastructure sharing among operators J. Kibiłda, N. Kaminski, and L. A. DaSilva, Radio Access Network and Spectrum Sharing in Mobile Networks: A Stochas:c Geometry Perspec:ve, IEEE Trans. on Wireless Communica:ons, vol. 16, no. 4, Apr J. Kibiłda, B. Galkin, and L. A. DaSilva, Modelling Mul:-Operator Base Sta:on Deployment Pacerns in Cellular Networks, IEEE Trans. on Mobile Compu:ng, vol. 15, no. 12, pp , Dec. 2016

8 The problem Problem: Mul:-tenancy, ac:ve sharing of radio access network and backhaul, and mul:ple modes of spectrum sharing, such as LSA and CBRS, are being explored by wireless network operators. We seek to understand the impact of sharing on coverage and capacity. State-of-the-Art: Single-operator stochas:c models of infrastructure deployment. Performance assessment using simula:on, drive tests. Our work: Develop a stochas:c model of mul:-operator network deployment. Employing stochas:c geometry, we derive results on the impact of spectrum and RAN sharing on coverage and capacity.

9 Models and reality of mul;-operator deployments

10 Spectrum and RAN sharing between MNOs Kibiłda, Kaminski, DaSilva [IEEE TWC 2017] ü No sharing ü Infrastructure only ü Spectrum only ü Infrastructure + Spectrum (Full)

11 Comparing these alterna;ves ü Metrics: ü Coverage ü Throughput ü Dependence on: ü Operator network deployment pacerns ü Sharing agreements in place ü Spectrum sharing coordina:on

12 Independent deployment by two operators Poisson point process (PPP) case ü For no sharing and full sharing cases, can rely on results derived in the literature ü For infrastructure only and spectrum only sharing, we can derive results for coverage and average data rate

13 Analy;cal expressions for coverage and user rate PPP with intensi:tes λ 1 and λ 2, with η =λ 1 /λ 2 Shared Infrastructure Coverage User rate Shared Spectrum* Coverage User rate *We differen:ate between sharing of spectrum with bands that experience flat and frequency-selec:ve fading

14 Valida;on We can cross-validate closed-forms and simula:ons ü Channel bonding ü Best channel selec:on

15 Independent PPP is a simplis;c assump;on Cluster processes to model mul:-operator deployments ü Premise: mul:-operator RAN deployments exhibit significantly more clustering than single-operator ü Inves:gated goodness of fit of log-gaussian Cox process (LGCP), Matern cluster process (MCP) and Thomas process (TP) ü Deployment data from Ireland, Poland, and the UK

16 Some of our results Kibiłda, Galkin, DaSilva [IEEE TMC 2016] ü Combined mul:-operator deployments cluster at shorter distances (high demand areas) and repulse at longer ü Log-Gauss Cox Process provides the closest match to real data ü Results are robust to several ci:es tested for in Europe

17 Coverage probability PPP LGCP GPP u 0 Spectrum sharing sceheme selec:on aggrega:on

18 And the conclusions so far ü Infrastructure and spectrum cannot be simply subs:tuted for each other, as they bring a tradeoff in coverage and throughput ü The efficiency gains from the combina:on of infrastructure and spectrum sharing do not add linearly ü The spa;al distribu;on of the networks has a significant impact on the gains brought about by sharing ü Channel aggrega;on brings about gains to data rate, while best spectrum selec;on yields significant improvement to coverage ü The loss of coverage due to channel aggrega:on may be avoided, without hur:ng the user rate, with spa;al interference coordination

19 Drones to augment coverage and capacity in urban areas B. Galkin, J. Kibiłda, and L. A. DaSilva, Coverage Analysis for Low-al:tude UAV Networks in Urban Environments, IEEE Globecom 2017 B. Galkin, J. Kibiłda, and L. A. DaSilva, Backhaul for Low-al:tude UAVs in Urban Environments, IEEE ICC 2018

20 The problem Problem: UAVs are being considered for a range of commercial and public safety applica:ons. They can also be used to augment operators assets and provide mobile coverage. State-of-the-Art: Op:mal posi:oning solu:ons for drones serving users on the ground. Ini:al stochas:c models of links from ground sta:ons to drones. Our work: Capture the performance of low-al:tude UAVs serving as access points in urban environments. Consider both link to ground users and backhaul requirements. Model co-channel interference and effects of building blockage.

21 Model Ground user coverage ü Modeling building loca:ons and heights following ITU-R recommenda:on P ü Building heights are a random variable ü LOS probability a func:on of horizontal and ver:cal distances between UAV and ground user

22 Analysis An outline ü Derive expressions for the probability density func:on of the distance between the ground user and the serving UAV ü Condi:oned on LOS and NLOS ü Condi:onal coverage probability can then be expressed in terms of the Laplace transform of the LOS and NLOS interference ü If UAV deployments follow a PPP, LOS and NLOS interfering UAVs form an inhomogeneous PPP ü Decondi:on to obtain the coverage probability

23 Coverage probability Density: 25/km 2 ; beamwidth: 2.84 rad

24 Coverage probability SINR threshold: 0 db; beamwidth: 2.84 rad

25 Model Backhaul

26 Backhaul probability LTE technology

27 Average supported data rates LTE technology

28 Backhaul probability mmwave technology

29 Backhaul probability as a func;on of GS height LTE technology

30 Backhaul probability as a func;on of GS height mmwave technology

31 Conclusions ü Low-al;tude UAVs serving ground users can benefit from the interference-mi:ga:ng effects of building blockage ü Sigmoid models of LOS based on ver:cal angle only may provide misleading results ü Good backhaul probability can be obtained with dedicated ground sta;ons deployed much less densely than exis:ng base sta:on infrastructure ü UAVs would benefit from direc;onal antennas with intelligent beam alignment, to compensate for LOS interference from distant GSs ü When using LTE for the backhaul, the op:mum ground sta:on height is consistent with exis:ng LTE infrastructure: poten;al for coloca;on

32 Acknowledgements ü Much of the work presented today is due to Conor Sexton, Boris Galkin and Jacek Kibiłda ü This work is supported by the Science Founda:on Ireland

33 luizdasilva.wordpress.com connectcentre.ie