A 5G Paradigm Based on Two-Tier Physical Network Architecture Elvino S. Sousa Jeffrey Skoll Professor in Computer Networks and Innovation University of Toronto Wireless Lab IEEE Toronto 5G Summit 2015 (Nov 14/2015) 1
Cellular System Evolution 1G, 2G, 3G, 4G, clear. 1 Analog, 2 - digital voice, 3 - voice plus variable data, 4 LTE-Advanced (Internet access). 1978 2012, or about 8 years/generation Computing: Similar approach to classification of generations (1G-4G, 1953 1982). 5 th generation?... Not clear and we could also loose interest in calling it a generation (at least from technology standpoint). One difference: Terminology here driven by industry group and standards. Currently we see the search is on for 5G, and it seems simultaneously for beyond 5G. IEEE Toronto 5G Summit 2015 (Nov 14/2015) 2
Classification by Generations Going forward it will be more like looking to the past, as opposed to 1-3G. 4G to some extent was coined after the fact. 4G defining features: OFDM on wider bandwidths, MIMO But MIMO not widely implemented because of terminal constraint IEEE Toronto 5G Summit 2015 (Nov 14/2015) 3
Key to Network Capacity Goal of wireless systems is to provide not only sufficient user capacity (in terms of capability of modulation scheme) but also network capacity Ultimately network capacity depends on More efficient physical layer, interference control Quantity of Spectrum Cell size (Small Cells) IEEE Toronto 5G Summit 2015 (Nov 14/2015) 4
My Take on Recent History My view of 4G (evolution to higher capacities beyond 3G) 2003 white paper Autonomous infrastructure wireless networks 4G is Here. 2007? Femto cells Later small cells. Reality: Data offloading - WiFi integrated into all smart phones IEEE Toronto 5G Summit 2015 (Nov 14/2015) 5
Future Goal Current data usage (typical plans) In home Internet access: Typical plan 90 GB/month Cellular typical plan: 3 GB/month Ratio: 30:1, or one month to one day. Goal: Make wireless access ( as accessible as wired access, i.e. 1:1. IEEE Toronto 5G Summit 2015 (Nov 14/2015) 6
1000x Capacity Common goal, referred to currently, is 1000x capacity Not achievable with improvements in modulation and even spectrum allocation. Solution: small cells Problem: Infrastructure to deploy small cells may be costly. IEEE Toronto 5G Summit 2015 (Nov 14/2015) 7
Some Architectural Approaches Femtocells Use existing network connection points Business issues (private, public femtocell) Third party traffic on network connection point? Small Cells Larger than Femtocells Infrastructure Cost Cloud RAN Best for Intercell-Interference control Rigidity in expansion Organic growth at physical layer? Back-haul/Front-Haul Approaches IEEE Toronto 5G Summit 2015 (Nov 14/2015) 8
Alternative to Small Cells Increase the efficiency of existing cells sites. Achieve large capacities by use of large antennas at the terminal. Space division multiplexing at the Terminal. IEEE Toronto 5G Summit 2015 (Nov 14/2015) 9
Two-Tier Networks Primary Nodes (P node): Classical Base station with Large Antenna and Power Capability. Secondary Nodes (S node): Transceiver with large capability for spatial processing. D1 Distance to P node. User Equipment (UE). Small, with usual antenna and power limitations. D2 Distance to S node. D1 >> D2. Tier 1 spectrum more precious depending on ratio D1/D2. IEEE Toronto 5G Summit 2015 (Nov 14/2015) 10
Two-Tier Network Architecture Tier 1 P node Tier 2 S node IEEE Toronto 5G Summit 2015 (Nov 14/2015) 11
Spectrum Tier 1 Licensed Tier 2 Licensed or Unlicensed? IEEE Toronto 5G Summit 2015 (Nov 14/2015) 12
Physical Layers Tier 1 Cellular Technologies, e.g. 4G, 5G, evolution. Key: Interference Management Tier 2 Cellular Technologies, WiFi and evolutions, LTE-U, Bluetooth, other technologies IEEE Toronto 5G Summit 2015 (Nov 14/2015) 13
S Node Sharing Options Dedicated to a user terminal. Active vs. Passive Shared among a user account. Shared among public users. IEEE Toronto 5G Summit 2015 (Nov 14/2015) 14
S Node Design Use Cases Generic Nodes based on Antenna Size, and or configurations, i.e. antenna design Fixed nodes for home? Fixed nodes for commercial applications? Nodes for vehicle (generic or specific to vehicle model). Notes for buses, street cars, trains? Portable organically deployed nodes, e.g. construction site. IEEE Toronto 5G Summit 2015 (Nov 14/2015) 15
Role for Millimeter Wave Spectrum Use mmwave in Tier 1 for Line-of-sight scenarios (use cases). S Node: Capability for both microwave and mmwave on Tier 1 interface, selfconfigurable (mmwave, microwave) to handle blocked propagation cases. IEEE Toronto 5G Summit 2015 (Nov 14/2015) 16
Public/Private Networks Tier 1 P nodes Public Slower to deploy Tier 2 S nodes Private or Public? Organic deployment IEEE Toronto 5G Summit 2015 (Nov 14/2015) 17
Physical Layer Design Tier 1 S node S node fixed Easy channel estimation (reference signals) Easier Power Control Easier resource allocation and antenna adaptation S node mobile (e.g. vehicle, or bus) Tier 2 - UE Predicable motion Advantages in channel estimation, power control, antenna adaptation, resource allocation Current approaches in smartphones (cellular & WiFi), other technologies IEEE Toronto 5G Summit 2015 (Nov 14/2015) 18
S Node Antenna Design Massive Beam-forming Massive MIMO Other spatially efficient technologies e.g. for mmwave. IEEE Toronto 5G Summit 2015 (Nov 14/2015) 19
Tier 1 Signal Design New Strategy for pilot signal design LTE pilot approach does not scale for large MIMO. When one antenna transmits, all the others are off! Signal design based on class of use cases. Different modes depending on use case and S node class. IEEE Toronto 5G Summit 2015 (Nov 14/2015) 20
Optimum Ratio of D2 to D1 UE has two options Connect directly to P node Connect through S node Strategy for selection of P node or S node Distance/power consideration Spectral efficiency (need D2 not too large) Traffic Load at P node. Latency issues? IEEE Toronto 5G Summit 2015 (Nov 14/2015) 21
Network Optimization: Self- Organizing S Node First time when the S node is turned on, it goes into initialization mode and self-configuration. Network environment learning Optimization of Tier 1: P node to S node link. In the initialization mode the S node senses the spectrum and chooses a band with minimum interference and if necessary it communicates with nearby co-channel S nodes to perform selfconfiguration. IEEE Toronto 5G Summit 2015 (Nov 14/2015) 22
Indoor Use Case? Indoor cells vs. Coverage from the outside. Two systems (macro/femto) or continuum? Do we isolate indoors? Could even modify construction Advantage of isolating indoors vs. Flexibility of capturing external signals IEEE Toronto 5G Summit 2015 (Nov 14/2015) 23
Vehicular Use Case? Vehicular movement is regular. Room for better smart antennas. Many possibilities here for a new system architecture. Antenna custom design to model. Initial benefits: (hot spot at night?, one wireless account?). Other better benefits. IEEE Toronto 5G Summit 2015 (Nov 14/2015) 24
Vehicular Scenarios Vehicle in the city high level MIMO link to P node or other highly directional antenna Vehicle at home user at home, evening and night Vehicle in the country side link range extension Vehicle in the highway (remote area) reduce density of base stations required to cover highways. IEEE Toronto 5G Summit 2015 (Nov 14/2015) 25
Summary Two approaches for large cellular network capacities Small Cells Need the associated infrastructure Ultimately achieves the highest capacities Two-Tiering Organic deployment of S nodes Antenna offloading Public/private S node public sharing vs. private Tier 2 technologies? IEEE Toronto 5G Summit 2015 (Nov 14/2015) 26