E401: Advanced Communication Theory · wireless communications I It has been incorporated into...
Transcript of E401: Advanced Communication Theory · wireless communications I It has been incorporated into...
E401: Advanced Communication Theory
Professor A. ManikasChair of Communications and Array Processing
Imperial College London
Multi-Antenna Wireless CommunicationsSpatiotemporal Wireless Communications, mmWave, Massive MIMO
& Distrubuted Antenna Arrays
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 1 / 126
Table of Contents1 Increasing the Degrees of Freedom 32 Massive MIMO 8
Introduction 9Massive MIMO (maMI): What and Why? 11maMI BS Antenna Geometries 14Summary of Main Advantages and Challenges 16
3 Spatiotemporal Wireless Communications 17Introduction 17Classification of Spatiotemporal Receivers 19Examples: Decoupled Space & Time CDMA Rx 21Spatiotemporal Manifolds (Extended Manifolds) 24DefinitionsThe "Shifting Matrix"A Generic-Rx Spatiotemporal Structure3D-data CubeA Representative Example: STAR Manifold Rx
Estimation Problem M>N 37Spatiotemporal Channel EstimationMain Properties of STAR subspace-type Receivers:
Reception Problem (Spatiotemporal Beamforming) 39STAR Array PatternExample: Space & Spatiotemporal Gain PatternsSpatiotemporal BeamformersSpatiotemporal Capacity
Spatiotemporal Representative Examples 48Example-1: ST-Array Pattern and ST-CapacityExample-2: Spatiotemporal Channel EstimationExample-3: Constellation DiagramExample 4: SNIRoutExample-5: M>N=2 - good for "handsets"Example-6: Spatiotemporal Beamformer versus Massive MIMO
4 mmWave Communications 60Introduction 61mmWave Spectrum Utilisation 62mmWave Communications versus LTE 6GHz 63mmWave MIMO: Main Characteristics 65mmWave Antenna Array Chipsets 68mmWave-MIMO: Main Advantages and Challenges 70
5 mmWave Beamformers 71Digital Beamformer 71Software Defined Radio Implementation 72Modelling a Single Branch of a Digital Beamformer 76Analogue mmWave Beamformer 82Hybrid mmWave Beamformer 85Comments 90Overview of Digital, Analogue & Hybrid Beamformers 91
6 Distributed Antenna Arrays (LAA) 92Introduction 93Wideband Assumption (WB-assumption) 94Wideband vs Narrowband assumption 95WideBand Assumption Signal Model 96Example 98
7 5G 100Introduction 101Revolutionary Changes 104High Capacity Requirements 114ITU Recommendations 119New Radio 1225G Frame Structure 123Notes 124
8 An Imaginary Illustration 126Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 2 / 126
Increasing the Degrees of Freedom
Increasing the Degrees-of-Freedom
There is an increased interest for beamformer-based communicationsin 5G. Issues:
I high path loss,I very dense co-channel interference environment, andI multipath effects in frequency selective channels.
Solutions :1 to employ massive MIMO ,2 to employ statiotemporal algorithms (manifold extenders)3 to employ both massive MIMO and manifold extenders
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 3 / 126
Increasing the Degrees of Freedom
Solution-1Solution: to employ massive MIMO , i.e. to increase the"degrees-of-freedom" by increasing N (remember: M < N)
I Multiple-antenna (MIMO): the technology is becoming mature forwireless communications
I It has been incorporated into wireless broadband standards like LTEand Wi-Fi.
I There are two cases:
F Non-parametric massive: The antennas are not working together butare independent units. This is a problematic approach(N=↑ ⇒number-of-unknowns=↑)
F Parametric massive: based on array processing (antennas are workingtogether as a single unit. This is a good approach(N=↑ ⇒number-of-unknowns 6=↑).
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 4 / 126
Increasing the Degrees of Freedom
Solution-2Solution: use array processing in conjunction with
I spatiotemporal algorithms. orI "virtual" arrays, orI both spatiotemporal algorithms and "virtual" arrays.
In this case we keep the number of antennas N fixed but we extendthe "array manifold" to
I spatiotemporal manifolds,I virtual array manifolds orI virtual-spatiotemporal manifolds
Solution-3Solution: we increase the number of antennas N and, at the sametime, we extend the array manifold (both Solutions 1 and 2)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 5 / 126
Increasing the Degrees of Freedom
Solution-1:
I massive SIMO Rx and massive MIMO Rx: M < NI Remember:
F M = number of users/sources/signals/transmitters,F N = number of Rx antennas
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 6 / 126
Increasing the Degrees of Freedom
Solutions 2 and 3:
I spatiotemporal Rx: M can be greater than N (M > N)I Remember that for space-only RX: M < N
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 7 / 126
Massive MIMO
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 8 / 126
Massive MIMO Introduction
Massive MIMO: Introduction
Massive MIMO also known as Large-Scale Antenna Systems, Very LargeMIMO, Hyper MIMO and Full-Dimension MIMO
Definition (MIMO and maMI)
MIMO (multiple input, multiple output) is an antenna arraytechnology for wireless systems (e.g. wireless comms or radar) whereantenna arrays are used at both the transmitter (source) and receiver(destination).
However, there is no a clear definition when a MIMO system isdefined as "massive MIMO" (maMI)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 9 / 126
Massive MIMO Introduction
some say:
Massive MIMO = Employing 100s of antennas at the base station.
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 10 / 126
Massive MIMO Massive MIMO (maMI): What and Why?
Massive MIMO: What and Why?maMI : makes a clean break with current practice through the use ofa very large number of antennas N (e.g., hundreds or thousands)that are operated fully coherently and adaptively.
I Extra antennas⇓focusing the transmission and reception of signal energy intoever-smaller regions of space⇓huge improvements in throughput and energy effi ciency
MaMI was originally envisioned for time division duplex (TDD)operation (channel estimation is carried out only on the uplink andthe estimated parameters are used in the downlink),
MaMI can potentially be applied also in frequency division duplex(FDD) operation (channel estimation can be carried out on bothuplink and downlink).
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 11 / 126
Massive MIMO Massive MIMO (maMI): What and Why?
The more antennas N the transmitter/receiver is equipped with,I the more the possible signal paths/users/sources/Tx (M) can be
F detected,F resolved andF their parameters can be estimated
and
I the better the performance in terms of data rate and link reliability.
The price to pay isI increased complexity of the hardware (number of RF amplifierfrontends) and
I the complexity (longer vectors and bigger matrices to handle) andenergy consumption of the signal processing at both ends.
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 12 / 126
Massive MIMO Massive MIMO (maMI): What and Why?
BenefitsOther benefits of massive MIMO include:
I the extensive use of inexpensive low-power components,I reduced latency ,I simplification of the media access control (MAC) layer, andI robustness to interference and intentional jamming.
ChallengesWhile massive MIMO renders many traditional research problemsirrelevant, it uncovers entirely new problems that urgently needattention; for example,
I the challenge of making many low-cost low-precision componentsto work effectively together,
I the need for new channel estimation approaches,I reducing internal power consumption to achieve total energyeffi ciency reductions
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 13 / 126
Massive MIMO maMI BS Antenna Geometries
maMI BS Antenna Geometries
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 14 / 126
Massive MIMO maMI BS Antenna Geometries
Example of Massive MIMO: Macro-cells
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 15 / 126
Massive MIMO Summary of Main Advantages and Challenges
Summary of Main Advantages and ChallengesmaMI Advantages
Increased data rate .Significant reduction in air-latency .Simplifies the MAC layer.Improved energy effi ciency .Improved interference suppression .
maMI Challenges
Handling huge amounts of data for baseband signal processing.Low cost RF chains required.Effi cient calibration and synchronisation across multiple RF chainsrequired.Internal power consumption constraints.Channel estimation is challenging utilising conventionalnon-parametric models.
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 16 / 126
Spatiotemporal Wireless Communications Introduction
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 17 / 126
Spatiotemporal Wireless Communications Introduction
Space-Time Communications
Space-Time Communications can be employed in TDMA/FDMA,CDMA, OFDMA and NR type systems
N.B.:I In TDMA/FDMA the signal is not spread and only few (mainly oneor two - i.e. M < N) strong cochannel interferences (CCI) are presentwhen the system employs channel reuse between cells.The array can be used to null (remove/reduce) these fewinterferences
I In CDMA all active users use the same bandwidth and are separatedby employing different PN-codes to reduce/remove the MAIinterference from other user.i.e. in a CDMA environment the array has to deal with a verylarge number of weak interferences (M > N).
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 18 / 126
Spatiotemporal Wireless Communications Classification of Spatiotemporal Receivers
Classification of Spatiotemporal Receivers
1 Decoupled Space and Time Rxs.∃ two main Rx architectures
1 "Space" and then "Time"
2 "Time" and then "Space"
2 STAR-Rx’s (SPATIOTEMPORAL ARRAY manifold Rx), where thetime and space are integrated.∃ one generic STAR-Rx architecturea
aN.B.: STAR should not to be confused with STAP (space-time adaptive processing)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 19 / 126
Spatiotemporal Wireless Communications Classification of Spatiotemporal Receivers
Definition ("Space" and then "Time" )
Definition ("Time" and then "Space" )
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 20 / 126
Spatiotemporal Wireless Communications Examples: Decoupled Space & Time CDMA Rx
Example (ST-CDMA Rx Classification)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 21 / 126
Spatiotemporal Wireless Communications Examples: Decoupled Space & Time CDMA Rx
Example (Decoupled Space & Time CDMA Receiver)
N.B.: each path is received/isolated individually (i.e. no diversity)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 22 / 126
Spatiotemporal Wireless Communications Examples: Decoupled Space & Time CDMA Rx
Example (Decoupled Space & Time RAKE CDMA Receiver)
N.B.: Multipath diversity (all multipaths are separated and combined)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 23 / 126
Spatiotemporal Wireless Communications Spatiotemporal Manifolds (Extended Manifolds)
Spatiotemporal Manifolds (Extended Manifolds)
array manifold: we can add more wireless parameters from the Tx, Rxand channel
For instance
S(θ, φ,Fc , c , r x , r y , r z ,pseudorandom sequ, delay, polarisation parameters,No.of subcarriers/carriers, bandwidth, Doppler frequency).⇓⇓h ,spatiotemporal manifold [it is a function of the
original S(θ, φ,Fc , c, r x , r y , r z )]
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 24 / 126
Spatiotemporal Wireless Communications Spatiotemporal Manifolds (Extended Manifolds)
Extended Manifolds (cont.)
Extension of the spatial manifold to the extended manifold
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 25 / 126
Spatiotemporal Wireless Communications Spatiotemporal Manifolds (Extended Manifolds)
Extended array manifold vectors can be re-expressed as a function of spatialmanifold vector S .
Examples
h = AS (1)
h = S ⊗ A (2)
h = (S ⊗ A)� B (3)
where h, S , A, B, A are functions of the parameters of interest
Definition
spatiotemporal manifold: H ,locus of all vectors h
Advantages:I No need to perform the same analysis multiple times but to express the analysisas a function of the "spatial" manifolds
I Easier to evaluate the effect of changing the system architecture
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 26 / 126
Spatiotemporal Wireless Communications Spatiotemporal Manifolds (Extended Manifolds)
ExamplesExtensions of the array manifold for use in modern digital communications
I Basic STAR manifoldhSTAR = S ⊗ J`c (4)
I Polar-STAR manifoldhPOL-STAR = S ⊗ q⊗ J`c (5)
I Doppler-STAR manifold
hDOP-STAR = S ⊗(
J`c �FD)
(6)
I Multicarrier-STAR manifold
hMC-STAR = S ⊗ J`α [`,Fk ] (7)
MIMO: the above extended maniolds but also include the Tx-array propertiesI e.g. virtual-STAR
hvirtual-STAR = S ⊗ S∗⊗ J`c (8)
"Functions of Array Manifolds", IEEE Transactions on Signal Processing, Vol59, Issue 7, p.3272-3287, 2011
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 27 / 126
Spatiotemporal Wireless Communications Spatiotemporal Manifolds (Extended Manifolds)
The "Shifting Matrix"
The matrix J is known as a shifting matrix (a Next ×Next matrix)defined as follows
J , JNext=
0 0 · · · 0 01 0 · · · 0 00 1 · · · 0 0....... . .
......
0 0 · · · 1 0
=[0TNext−1 0INext−1 0Next−1
](9)
having the property that every time the matrix J (or JT ) operates ona column vector it down-shifts (or up-shifts) the elements of thevector by one.
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 28 / 126
Spatiotemporal Wireless Communications Spatiotemporal Manifolds (Extended Manifolds)
N.B.: J`x is a version of x down-shifted by ` elements, while(JT)`x is a version of x up-shifted by ` elements.
Examples
x =
x1x2x3...xK
; J3x =
000x1x2...
xK−3
;(JT)2x =
x3x4...xK00
(10)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 29 / 126
Spatiotemporal Wireless Communications Spatiotemporal Manifolds (Extended Manifolds)
A Generic Spatiotemporal Rx Architecture
· at point A: x(t) =function{S}+n(t)· at point B: x [n] =function{h}+n[n]
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 30 / 126
Spatiotemporal Wireless Communications Spatiotemporal Manifolds (Extended Manifolds)
3D-data Cube
x [n] = vec{
X[n]T}
(11)
= function{h}+ n[n] (12)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 31 / 126
Spatiotemporal Wireless Communications Spatiotemporal Manifolds (Extended Manifolds)
Example (A Representative Rx: STAR Manifold Rx)Let us focus on an extended manifold which is:S(θ, φ,Fc , c , r x , r y , r z ,pseudorandom sequ, delay).⇓h = basic STAR= S ⊗ J`c (see Equation 4)
The above is suitable for many antenna array systems, including CDMAsystems.For the j th path of the i th user the array manifold vector is
S(θij , φij ) = exp(−j [r1, r2, . . . , rN ]T k(θij , φij ) ∈ CN×1
where [r1, r2, . . . , rN ] represents the array geometry and k(θij , φij ) is thewavenumber vector.In this case, the extended manifold is
hij4= S ij ⊗ Jlij ci ∈ CNNext×1 (13)
with S ij , S(θij , φij ) and hij , h(θij , φij , `ij )
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 32 / 126
Spatiotemporal Wireless Communications Spatiotemporal Manifolds (Extended Manifolds)
Example (cont.)To achieve this extension, the array received signal vectorx (t) ∈ CN×1 is transformed to a discretised signal x [n] ∈ CNNext×1
Next = 2× q×Nc , where: Nc=code period; q =oversampling factor.
Note: here we will use q = 1. That is, the array received signal vectorx (t) is discretised by a chip rate sampler, i.e. Ts = Tc .
The parameter ` is the discrete delay: ` =⌈
τTc
⌉modNc .
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 33 / 126
Spatiotemporal Wireless Communications Spatiotemporal Manifolds (Extended Manifolds)
Example (cont.)
The discrete samples are then passed through a "manifold extender" (in thisexample this is a tapped-delay line of lengtha equal to 2Nc ).This is to ensure that one whole data symbol of the desired user and thecorresponding multipath components are captured within this 2Nc interval.
athis may may also of length Nc to reduce the ISI (Inter-Symbol Interference)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 34 / 126
Spatiotemporal Wireless Communications Spatiotemporal Manifolds (Extended Manifolds)
Example (cont.)As shown in preprocessor’s figure the received space-time signalvector x [n] is formed by concatenating the contents of thetapped-delay lines of all the antennas,
i.e. x [n] =[x1 [n]
T , x2 [n]T , ... , xN [n]
T]T∈ C2NcN×1 (14)
= vec(X[n]T ) (15)
where xk [n] represents the contents of the tapped-delay line at thek th antenna associated with the nth data symbol period, and
Rxx = E{x [n] .x [n]H
}∈ C2NcN×2NcN (16)
Note that the whole theory presented in the previous TOPIC "ArrayReceivers for SIMO and MIMO", can be applied directly here usingthe above Rxx
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 35 / 126
Spatiotemporal Wireless Communications Spatiotemporal Manifolds (Extended Manifolds)
Example (cont.)
For M users (assuming the 1st to be the desired user) and afrequency selective channel of Ki paths for the i-th user, the vectorx [n] ∈ CNNext×1 can be expressed as follows:
x [n] =M
∑i=1[Hi ,prevβi
, Hi βi, Hi ,nextβi
]
ai [n− 1]ai [n]
ai [n+ 1]
+ n[n] (17)x [n] = H1m1[n] + I ISI [n] + IMAI [n] + n[n] (18)
where
H1 =[h11, h12, ..., h1K1
](19)
IISI[n] =
(IN ⊗
(JT)Nc)
H1m1[n-1] +(
IN ⊗ JNc)
H1m1[n+1]
(20)
m1[n] =
{a[n].β
1for slow fading
a[n].β1[n] for fast fading
∈ CK1×1 (21)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 36 / 126
Spatiotemporal Wireless Communications Estimation Problem M>N
Estimation Problem M>NSpatiotemporal Channel Estimation
The receiver initially estimates, over an observation time, thespatio-temporal manifold parameters of the desired signal(s), whichare then employed to remove the MAI and ISI termsThe estimation, for instance, can be carried out using the following2D. ‘STAR’subspace cost function (n-th interval):
ξ(θ, `) =1
h(θ, `)HPn h(θ, `)(22)
In Equation 22 the matrix Pn is the projection operator associatedwith the "noise subspace" of
Rxx = E{x [n] .x [n]H
}i.e. this projection operator should be estimated from the 2nd orderstatistics (i.e. Rxx ) of the x [n] ∈ CNNext×1 given by Equation 18.
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 37 / 126
Spatiotemporal Wireless Communications Estimation Problem M>N
Main Properties of STAR subspace-type CDMA Receivers:1 blind (estimation of channel parameters without pilots)
2 separates/estimates all the paths of the desired user in the presenceof MAI
3 The number of multipaths that can be resolved is not constrained bythe number of array elements (antennas), i.e.
M > N
4 near-far resistant (i.e. in CDMA there is no need for power control)
5 superresolution capabilities.
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 38 / 126
Spatiotemporal Wireless Communications Reception Problem (Spatiotemporal Beamforming)
The Reception ProblemDefinition (STAR Array Pattern)If the array elements are weighted by complex-weights then the array patternprovides the gain of the array as a function of DOAs
if (θ,`) 7−→ h (θ, `) then g(θ, `) = wHh(θ, `) (23)
where g(θ, `) denotes the STAR gain of the array for a signal arriving fromdirection θ and delayed by τ
with ` ,⌈
τ
Tc
⌉modNc (24)
then the function g(θ, `), ∀θ and ∀`, is known as the STAR Array Pattern
N.B.: default pattern :I space only:g(θ, `) = 1T2NNc h (θ, `), i.e. w = 12NNc (i.e. no weights)
I spatiotemporal:g(θ, `) = (1N ⊗ c)T h (θ, `), i.e. w = h (90◦, 0Tc ) , (i.e. no weights)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 39 / 126
Spatiotemporal Wireless Communications Reception Problem (Spatiotemporal Beamforming)
Example: Space & Spatiotemporal Gain Patterns
Consider a Uniform Linear Array of N elements using a PN-code oflength Nc = 31.
Examples of the STAR array pattern forI delays: 0 and 7TcI directions: 90◦ and 120◦
and N = 2, 3, 4, 5 are given below:
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 40 / 126
Spatiotemporal Wireless Communications Reception Problem (Spatiotemporal Beamforming)
Example (N = 2)
w = 12NNc w = h(90◦, 0Tc ) w = h(120◦, 7Tc )
Example (N = 3)
w = 12NNc w = h(90◦, 0Tc ) w = h(120◦, 7Tc )
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 41 / 126
Spatiotemporal Wireless Communications Reception Problem (Spatiotemporal Beamforming)
Example (N = 4)
w = 12NNc w = h(90◦, 0Tc ) w = h(120◦, 7Tc )
Example (N = 5)
w = 12NNc w = h(90◦, 0Tc ) w = h(120◦, 7Tc )
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 42 / 126
Spatiotemporal Wireless Communications Reception Problem (Spatiotemporal Beamforming)
Example (focusing on the pattern for N = 5)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 43 / 126
Spatiotemporal Wireless Communications Reception Problem (Spatiotemporal Beamforming)
Spatiotemporal Beamformers
If a single-user Rx (SU-Rx) is used then we only need to estimate thechannel parameters of the i-th user only (with reference to the lastfigure, these are provided at the output of "Processor-2". Then, theSTAR manifold matrix of the i-th user can be formed in the block"Processor-3" as follows
Hi =[hi1, hi2, ..., hiLp
]∈ C 2NN c×Lp (25)
where Lp is the number of multipaths.
If a multi-user Rx (MU-Rx) is used then the channel estimator("Processor-2") should provide estimates for all users and thus Equ.25 should be formed in the block "Processor-3" for every i (i.e. ∀i).
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 44 / 126
Spatiotemporal Wireless Communications Reception Problem (Spatiotemporal Beamforming)
The concept of "spatiotemporal (STAR) weight vectors" can beextended to other receivers too.The following two receivers may be used to receive the multipathsignals (multipath diversity) of the i-th user.
STAR-RAKE (SU)
w i = Hi βi(26)
STAR-Subspace (SU)
w i = constant×PnHi
(HHi PnHi
)−1βi. (27)
whereI a scalar constant is used as a normalising factor such that ‖w i‖ = 1,and
I Pn is the projection operator associated with the "noise subspace" of
Rxx = E{x [n] .x [n]H
}Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 45 / 126
Spatiotemporal Wireless Communications Reception Problem (Spatiotemporal Beamforming)
Spatiotemporal Capacity
SISO capacity :C = B log2(1+ SNIRout ) bits/sec (28)
MIMO Capacity :
C = B log2
(det(Rxx )
det(Rnn)
)bits/sec (29)
or,equivalently,
C = B log2
(det(
IN +1
σ2nSRmmSH
))bits/sec (30)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 46 / 126
Spatiotemporal Wireless Communications Reception Problem (Spatiotemporal Beamforming)
If bandwidth −→ ∞ then C =?
SISO : limB→∞
C = 1.44Ps
N0 +NJ(31)
space SIMO : limB→∞
C = N × 1.44 PsN0 +NJ ↓
0
(32)
MIMO - virtual SIMO : limB→∞
C = N ×N × 1.44 PsN0 +NJ ↓
0
(33)
spatiotemporal-SIMO : limB→∞
C = N ×Next × 1.44Ps
N0 +NJ ↓0
(34)
virtual-spatiotemporal-SIMO : limB→∞
C = N ×N ×Next × 1.44Ps
N0 +NJ ↓0
(35)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 47 / 126
Spatiotemporal Wireless Communications Spatiotemporal Representative Examples
Spatiotemporal Representative ExamplesThe following examples are based on the following generic ST-Rx architecture and references
"STAR Channel Estimation in DS-CDMA Communication Systems", IEE Proc.-Commun.,
Vol. 151, No. 4, August 2004.
"Spatiotemporal-MIMO Channel Estimator and Beamformer for 5G", IEEE Trans. on
Wireless Comms, Vol. 15, No. 12, December 2016.Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 48 / 126
Spatiotemporal Wireless Communications Spatiotemporal Representative Examples
Examples (1. ST-Array Pattern and ST-Capacity)
limB→∞
C = NNext × 1.44Ps
N0 +Nj ↓
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 49 / 126
Spatiotemporal Wireless Communications Spatiotemporal Representative Examples
Examples (1. cont.)N = 5 antennas
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 50 / 126
Spatiotemporal Wireless Communications Spatiotemporal Representative Examples
Examples (2. Spatiotemporal Channel Estimation)3 users, 5 paths per user, 1st user=desired
User 1 (Desired) Path 1 Path 2 Path 3 Path 4 Path 5
Path Delay (Tc ) 1 9 17 21 27Path Direction (o) 50 94 125 141 76Path Coeffi cient -0.10 + 0.26j -0.01 - 0.24j -0.31 - 0.02j -0.31 - 0.02j 0.42 - 0.35jUser 2 (Interfer) Path 1 Path 2 Path 3 Path 4 Path 5
Path Delay (Tc ) 4 8 17 26 27Path Direction (o) 92 35 149 67 61Path Coeffi cient -0.20 + 0.56j -0.41 - 0.74j -0.39 - 0.92j -0.91 - 0.12j 0.76 - 0.00jUser 3 (Interfer) Path 1 Path 2 Path 3 Path 4 Path 5
Path Delay (Tc ) 2 13 19 25 27Path Direction (o) 103 84 80 79 116Path Coeffi cient -0.15 + 0.27j -0.71 - 0.24j -0.11 - 0.01j -0.21 - 0.05j 0.45 - 0.55j
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 51 / 126
Spatiotemporal Wireless Communications Spatiotemporal Representative Examples
Examples (2. cont.)
Surface and contour plots of the cost function (Equ 22) shows thatall 5 path delays and directions are correctly estimated
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 52 / 126
Spatiotemporal Wireless Communications Spatiotemporal Representative Examples
Examples (3. Constellation Diagram (Decision Variables))
1.5 1 0.5 0 0.5 1 1.51.5
1
0.5
0
0.5
1
1.5
Re
Im
2000 1500 1000 500 0 500 1000 1500 20002000
1500
1000
500
0
500
1000
1500
2000
Im
Re
ST Decorrel. MU Rx ST Decorrel. MU Rx (Incomp)
8 6 4 2 0 2 4 6 8
x 106
8
6
4
2
0
2
4
6
8x 106
Re
Im
STAR manifold Rx ST-RAKE RxProf. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 53 / 126
Spatiotemporal Wireless Communications Spatiotemporal Representative Examples
Examples (4. SNIR output)
SNIRout=f{M} ‘Near-Far’Resistance
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 54 / 126
Spatiotemporal Wireless Communications Spatiotemporal Representative Examples
Examples (5. M > N = 2: good for "handsets")
Desired user’s parameters = 8 paths with (TOA in Tc , DOA in degrees) asfollows:
I (5, 100◦), (6, 60◦), (12, 27◦), (15, 85◦),I (18, 45◦), (24, 30◦), (27, 118◦), (29, 105◦)
The desired user’s STAR Subspace-type spectrum (and contour diagram) foran array of N = 2 antennas, operating in the presence of three CDMA users isshown below:
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 55 / 126
Spatiotemporal Wireless Communications Spatiotemporal Representative Examples
Examples (6. Spatiotemporal Beamformer versus maMI (parametric))
see reference [2], page-48
Rx Antenna arraysI Circular Array of N=9 antennas: Spatiotemporal beamformer(16x9).
I Circular Array of N=500 antennas: Massive MIMO (maMI, 16x500)
4 co-channel users with 3 paths per user in a frequencyselective channel
the spatiotemporal manifold incorporates the following systemparameters: 5 subcarriers and a pn-code of length 15.
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 56 / 126
Spatiotemporal Wireless Communications Spatiotemporal Representative Examples
Examples (6. cont.)
N=9 antennas N=500 antennas
Multipath 1gain = 1334
Multipath 2gain = 1333
Multipath 3gain = 1336
Arr
ayG
ain
Delay
( sec)Ts
DOA(deg)A
rray
Gain
Multipath 1gain = 488
Multipath 2gain = 499
Multipath 3gain = 490
Delay
( sec)TsDOA(deg)
(a) Beampattern of a DopplerSTARsubspacereceiver consisting of 9 antenna elements.
(b) Beampattern of a massive MIMO subspacereceiver consisting of 500 antenna elements.
The 9-antenna spatiotemporal beamformer providesI higher gain and spatiotemporal selectivityI better performance (≈ 15dB, see paper)
than a traditional 500-antenna MIMO system
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 57 / 126
Spatiotemporal Wireless Communications Spatiotemporal Representative Examples
Examples (6. cont.)
N=500 antennas (spatiotemporal maMI)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 58 / 126
Spatiotemporal Wireless Communications Spatiotemporal Representative Examples
Examples (6. cont.)
N=9 antennas N=500 antennas
Beam 1gain = 1393
(Multipath 1)
Beam 2gain = 1438
(Multipath 2)
Beam 2 & 3(Multipath 2 & 3 collapse
into a single peak andare indistinguishable)
Beam 3gain = 1438
(Multipath 3)
Beam 1gain = 489
(Multipath 1)
Arr
ayG
ain
Delay( s sec)T DOA
(deg)
Delay( s sec)T DOA
(deg)
Arr
ayG
ain
(a) Beampattern of a DopplerSTARsubspacereceiver consisting of 9 antenna elements.
(b) Beampattern of a massive MIMO subspacereceiver consisting of 500 antenna elements.
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 59 / 126
mmWave Communications
mmWave Wireless Comms
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 60 / 126
mmWave Communications Introduction
mmWave Communications: Introduction
mmWave : milimeter wave or milimeter bandalso known by the ITU (International Telecommunications Union) asExtermely High Frequency (EHF) band.
Definition (mmWave)
mmWave,Spectrum 30GHz-300GHz ⇔short wavelegths from 10mmto 1mm!
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 61 / 126
mmWave Communications mmWave Spectrum Utilisation
mmWave Spectrum Utilisation
un-utilised spectrum: 30-60GHz. For instance,I 60GHz can be used for short range data linksI the IEEE WiFi standard 802.11ad run on 60GHz mmWave band
licenced spectrum (from FCC):I 71GHz-76GHz,I 81GHz-86GHz andI 92GHz-95GHz
for point-to-point high bandwidth communication links
mmWave can be used for high-speed (up to 10Gbps) wirelessbroadband communications.
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 62 / 126
mmWave Communications mmWave Communications versus LTE 6GHz
mmWave Communications versus LTE 6GHz
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 63 / 126
mmWave Communications mmWave Communications versus LTE 6GHz
Example (A Planar Array Geometry of 256 antennas at 30GHz)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 64 / 126
mmWave Communications mmWave MIMO: Main Characteristics
mmWave MIMO: Main Characteristics1. Very high atmospheric attenuation
I Rain attenuation (rain and humidity impact performance)I Atmospheric and molecular absorption (but only at certain frequencies)
This implies Huge propagation losses ⇒ strength=↓⇒range=↓i.e. limit the range of mmWave comms and mandate the requirementof algorithms that are powerful enough to combat these losses.
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 65 / 126
mmWave Communications mmWave MIMO: Main Characteristics
2. Very high path loss (free-space Friis attenuation ↑ with frequency).Thus, due to very small wavelength (milimeter) we observe:
I blockage (e.g. by objects such trees, buildings, etc.I diffractions.
This implies Sensitivity to Blockage, i.e due to the small wavelength, the
links in the 60 GHz band are very sensitive to blockage.
Examplesblockage by a human causes a 20-30 dB dip in the link budget.
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 66 / 126
mmWave Communications mmWave MIMO: Main Characteristics
3. Solution to very high atmospheric attenuation and path loss:I Highly directional beams. With current technologies, the size of theantenna element shrinks dramatically at these frequencies. This makesit possible to have a massive antenna array on the Tx and Rx and,thus, employ powerful beamforming algorithms at the Tx and Rx toobtain accurate and highly directional beams.
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 67 / 126
mmWave Communications mmWave Antenna Array Chipsets
mmWave Antenna Array Chipsets
Example (60GHz Antenna Array Chipset (Qualcomm))
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 68 / 126
mmWave Communications mmWave Antenna Array Chipsets
Example (Chipsets of Antenna Arrays with 8 Elements)
(a) 4x4 RF chips (brown squares), each with 8 antennas (yellow squares)
(b) possible distribution of 4x4 RF chips, each with 8 antennas mountednearby on a circuit board on a different substrate
(c) clock diagram for a single chip, including Tx and Rx chains, sharingantennas around the chip
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 69 / 126
mmWave Communications mmWave-MIMO: Main Advantages and Challenges
mmWave MIMO: Main Advantages and Challenges
Advantages
Possible to achieve very high data rates .Highly directional beams may be constructed leading to superiorinterference suppression.
Relieves the demand for spectrum .
Challenges
The channel presents very high level of fading and penetration loss—to the extent that rain can completely cut off a Tx signal!
Very little is known about the channel at these frequencies.
Handling huge amounts of data for baseband signal processing.
Low cost RF chains required - analog and hybrid beamformingneeds to be explored.
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 70 / 126
mmWave Beamformers Digital Beamformer
mmWave BeamformersDigital Beamformers
In a digital beamformer, each antenna element has its owncorresponding baseband port which offers the largest flexibility.
However, ADCs and DACs operating at multi-GHz sampling rates arevery power consuming; a full digital beamformer with several hundredantenna elements is very power hungry and complex.
Therefore early mmWave communication systems are expected to useanalog or hybrid beamforming architectures.
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 71 / 126
mmWave Beamformers Software Defined Radio Implementation
Software Defined Radio ImplementationRadio HW impementation (e.g mixers, filters, amplifiers, modulators,demodulators, detectors, etc), nowdays are implemented in softwareBasic MIMO System Architecture and SDR
↑DSP (e.g. RuspberryPie,PC)
↑SDR (e.g. FunCube Dongle, LimeDSR-mini)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 72 / 126
mmWave Beamformers Software Defined Radio Implementation
Digital mmmWave Beamformer ImplementationDigital Beamformer Digital BF PC-based
LO Local Oscillator DDC Digital Down Coverter
CLK Clock DDS Direct Digital Synthesiser
A/D Analogue to Digital Converter
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 73 / 126
mmWave Beamformers Software Defined Radio Implementation
SDRThe best SDR would have a high speed (and expensive) ADC for Rxand a high speed DAC for Tx at RF frequency.
1st Generation SDR Rx:
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 74 / 126
mmWave Beamformers Software Defined Radio Implementation
2nd Generation SDR Rx:
3rd Generation SDR Rx:
SDR Tx:
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 75 / 126
mmWave Beamformers Modelling a Single Branch of a Digital Beamformer
Modelling a Single Branch of a Digital Beamformer
A = k-th antenna = m(t) cos(2πFRF (t −
rTk k ic)
)= m(t) cos (2πFRF t − ψk )
where ψk , 2πFRFrTk k ic;FRF = Fc = carrier frequ.
B = 2 cos(2πFLO t) where FLO = FRF − FIFProf. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 76 / 126
mmWave Beamformers Modelling a Single Branch of a Digital Beamformer
C = A × B= 2m(t) cos (2πFRF t − ψk ) cos(2πFLO t)
= m(t) cos (2π (FRF + FLO ) t − ψk )
+m(t) cos (2π(FRF − FLO︸ ︷︷ ︸)t,FIF
− ψk )
D = m(t) cos (2πFIF t − ψk )
E = m(t`) cos (2πFIF t` − ψk )
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 77 / 126
mmWave Beamformers Modelling a Single Branch of a Digital Beamformer
Inphase:
I = 2 cos (2πFDLO t`)
I1 = m (t`) cos (2π(FIF + FDLO )t` − ψk ) +m (t`) cosψkI2 = m (t`) cos (ψk )
Quadrature:
Q = 2 sin (2πFDLO t`)
Q1 = m (t`) sin (2π(FIF + FDLO )t` − ψk ) +m (t`) sinψkQ2 = m (t`) sinψk
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 78 / 126
mmWave Beamformers Modelling a Single Branch of a Digital Beamformer
Baseband input signal (k-th antenna):
xk (t`) = I2 + j Q2
= m (t`) cosψk + jm (t`) sinψk = m (t`) exp(j 2πFRF
c rTk k i)
︸ ︷︷ ︸k th element ofmanifold vector
.
Note: FDLO = FIF =intermediate/digital-LO frequ.
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 79 / 126
mmWave Beamformers Modelling a Single Branch of a Digital Beamformer
Digital Tx-Beamforming
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 80 / 126
mmWave Beamformers Modelling a Single Branch of a Digital Beamformer
Digital Rx-Beamforming
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 81 / 126
mmWave Beamformers Analogue mmWave Beamformer
Analogue mmWave Beamformers
In analogue beamforming, one baseband port feeds an analoguebeamforming network where the beamforming weights are appliedeither directly on the analog baseband components, at someintermediate frequency, or at RF. For example, an RF beamformingnetwork may consist of several phase shifters, one per antennaelement, and optionally also variable gain amplifiers.
In any case, an analogue beamforming network typically generatesphysical beams but cannot generate a complex beam pattern.Especially in a multi-user environment this can lead to interference, ifpure beam separation is not suffi cient.
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 82 / 126
mmWave Beamformers Analogue mmWave Beamformer
Analogue Tx-Beamforming
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 83 / 126
mmWave Beamformers Analogue mmWave Beamformer
Analogue Rx-Beamforming
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 84 / 126
mmWave Beamformers Hybrid mmWave Beamformer
Hybrid beamforming, is a compromise between Digital andAnalogue Beamformers where a digital beamformer operating on afew baseband ports is followed by an analog beamforming network.
This architecture enables a compromise with respect to bothcomplexity and flexibility between analogue and full digitalbeamformer
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 85 / 126
mmWave Beamformers Hybrid mmWave Beamformer
Hybrid Tx-Beamforming
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 86 / 126
mmWave Beamformers Hybrid mmWave Beamformer
Hybrid Tx-Beamforming (vector representation)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 87 / 126
mmWave Beamformers Hybrid mmWave Beamformer
Hybrid Rx-Beamforming
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 88 / 126
mmWave Beamformers Hybrid mmWave Beamformer
Hybrid Rx-Beamforming (vector representation)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 89 / 126
mmWave Beamformers Comments
Comments
A beamforming receiver provides spatial selectivity (i.e. high gainreception of signals in desired directions while suppressingsignals in other directions ).Each individual antenna element, however, does not provide muchspatial selectivity. For a digital beamforming receiver this means thateach signal path, extending from respective antenna element all theway to the baseband port, will have to accommodate desired as wellas undesired signals.
Thus, to handle strong undesired signals, requirements on dynamicrange will be high for all blocks in the signal path and that willhave a corresponding impact on power consumption . In an analogbeamformer, however, the beamforming may be carried out already atRF and thus all subsequent blocks will need less dynamic rangecompared to the digital beamforming receiver.
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 90 / 126
mmWave Beamformers Overview of Digital, Analogue & Hybrid Beamformers
Digital BF Hybrid BF Analogue BFTX/RX weights at base-band (i.e. digital)
TX/Rx weights atboth RF (analogue)and baseband (digi-tal)
Tx/Rx weights at RFto form beams
Each antenna element(or antenna port) hasa transceiver unit (i.e.overall, high number oftransceiver units)
Each RF-beam has atransceiver unit (i.e.moderate number oftransceivers)
One transceiver unitand one RF-beam
good for frequency se-lective channels (onebeam per path)
conbination of digital(baseband) and ana-logue (RF) BF.
good for a frequencyflat channel
best for capacity andflexibility (high powerconsumption and cost -especially if BW=↑)
optim. for both cov-erage and capacity
best for coverage thanpower consumptionand cost
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 91 / 126
Distributed Antenna Arrays (LAA)
Distributed Antenna Arrays (or Large Aperture Arrays)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 92 / 126
Distributed Antenna Arrays (LAA) Introduction
IntroductionClassification:
I small aperture arraysI large/wide aperture or distributed arrays (many applications,e.g.repetitive localisation for trajectory tracking of mobile signals)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 93 / 126
Distributed Antenna Arrays (LAA) Wideband Assumption (WB-assumption)
Wideband Assumption (WB-assumption)Definition (Narrowband Assumption)If the transmitted baseband wavefront does not change while travellingacross the sensors of an array then different sensors see the same part ofthe transmitted signal and this is defined as the “narowband-assumption”(NB-assumption)
Definition (Wideband Assumption)If the transmitted baseband wavefront changes while travelling across thesensors of an array then different sensors see different parts of thetransmitted signal and this is defined as the “wideband-assumption”(WB-assumption)
N.B.:this should not be confused with the term “wideband signals”
If the array elements are distributed in space with large inter-sensorspacings, the WB-assumption is essential.
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 94 / 126
Distributed Antenna Arrays (LAA) Wideband vs Narrowband assumption
Wideband vs Narrowband assumptionWB-assumption: It is a function of the array geometry , sourcelocation and signal bandwidth
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 95 / 126
Distributed Antenna Arrays (LAA) WideBand Assumption Signal Model
WideBand Assumption Signal ModelNB-assumption and WB-assumption for M sources:
NB-assumption: x (t) =M
∑i=1S imi (t) + n(t) (36)
WB-assumption: x (t) =M
∑i=1S i �mi (t) + n(t), (37)
where, the “spherical wave manifold vector” is given by
S i = S(θi , ρi , array geometry, carrier freq)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 96 / 126
Distributed Antenna Arrays (LAA) WideBand Assumption Signal Model
Covariance MatrixI NB-assumption:
Rxx =M
∑i=1
PiS iSHi +Rnn (38)
I WB-assumption:
Rxx =M
∑i=1
S iSHi �Rmimi +Rnn (39)
whereRmimi = E
{mi (t)mi (t)
H}
(40)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 97 / 126
Distributed Antenna Arrays (LAA) Example
Example (Multi-source Trajectory Tracking)In the reference belowa, both the Wide-Band and Narrow-Bandassumptions are used for an array of sensors - where the sensors aredistributed in space,
the targets/sources parameters (kinematics of the targets) included inthe modelling are:
I ranges,I directions,I velocities and associated Doppler effects.
a
"Multi-source Spatiotemporal Tracking using Sparse Large Aperture Arrays",IEEETrans on Aerospace and Electronic Systems, pp.837-853, 2017
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 98 / 126
Distributed Antenna Arrays (LAA) Example
Example (cont.)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 99 / 126
5G
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 100 / 126
5G Introduction
Introduction
Definition5G is the 5th generation of mobile networks.
This is a significant evolution of todays 4G LTE networks. There arealso those who believe that this not simply an "evolution" but a"revolution" (4th industrial revolution, 5G ecosystem).
5G has been designed to meet the very large growth in data andconnectivity of today’s modern society, the internet of things withbillions of connected devices, and tomorrow’s innovations.
5G will initially operate in conjunction with existing 4G networksbefore evolving to fully standalone networks in subsequent releasesand coverage expansions.
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 101 / 126
5G Introduction
5G Ecosystem
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 102 / 126
5G Introduction
Definitions1 Femtocell: A femtocell is a small, low-power cellular base station,typically designed for use in a home or small business. A broader termwhich is more widespread in the industry is small cell, with femtocellas a subset. It is also called femto AccessPoint (AP).
2 Fantom Cell: (new concept in LTE-B)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 103 / 126
5G Revolutionary Changes
Revolutionary Changes: Latency
In addition to delivering faster connections and greater capacity, avery important advantage of 5G is the fast response time referred toas latency.
Latency is the time taken for devices to respond to each other overthe wireless network.
Examples (Latency)3G networks had a typical response time of 100 milliseconds,
4G is around 30 milliseconds and
5G will be as low as 1 millisecond(this is virtually instantaneous - opening up a new world of connectedapplications)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 104 / 126
5G Revolutionary Changes
5G Latency
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 105 / 126
5G Revolutionary Changes
LTE vs 5G Latency
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 106 / 126
5G Revolutionary Changes
Revolutionary Changes: New Comm. FamiliesEnhanced Mobile Broadband communications (emBC) — i.e.mobile communication links of very high bandwidth (broadband) thusdelivering very high speeds per user (multi Gbits/sec). For instance,mobile users with UHD screens need this family of comm. links.Massive Machine Communications (mMC) , - i.e. simultaneouscomm links to potentially millions of machines (e.g. smart cities).Mission Critical Communications (mCC) — i.e. communicationlinks of high reliability and low latency (e.g. comm links forself-driving cars or aerospace industry)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 107 / 126
5G Revolutionary Changes
5G Use Examples (Impact Areas)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 108 / 126
5G Revolutionary Changes
Other Revolutionary Changes
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 109 / 126
5G Revolutionary Changes
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 110 / 126
5G Revolutionary Changes
5G OperatesI on higher frequency bands, which offers significantly more availablebandwidth,
I on smaller cells, running at lower power and allowing networkdensification.
Uses MIMO antenna schemes (maximizing the reuse of scarcebandwidth). With massive MIMO beamforming, it becomes possibleto move the network forward from the traditional point-to-multipointparadigm to a real-time adaptive point-to-point link, with the basestation tracking the user and steering its signal to them.
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 111 / 126
5G Revolutionary Changes
Typical massive MIMO designs will range from tens to hundreds ofantennas, beamforming in azimuth and elevation.
Improved antenna gain overcomes path loss, enabling higher datarates per hertz compared with an isotropic signal at the same power.
mmWave in 5G:1 (+) directional communications ⇒⇓⇓
F (-) risk of deadnessF (-) Need of beam alighment/tracking to maintain connectivity(beamwidth vs tracking overheads trade-off)
2 (+) Massive MIMO ⇒ (−)Power consumption and device complexity
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 112 / 126
5G Revolutionary Changes
Summary of 5G New Characteristics
1 Flexibility for expanding connectivity needs
2 multi-connectivity across spectrum bands andtechnologies⇒improves coverage and mobility
3 Diverse spectrum types and bands (narrowband, wideband, UWB,TDD, FDD, exclusive use, shared use, or shared exclusive use )
4 A new unified air-interface provides the foundation (see NR)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 113 / 126
5G High Capacity Requirements
High Capacity Requirements
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 114 / 126
5G High Capacity Requirements
1. Improve Spectrum Effi ciency:A very large antenna array at each Base Station (BS) servingsimultaneoulsy a large number of co-channel users (massive MIMO),orA spatiotemporal array MIMO, orA spatiotemporal array massive MIMO
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 115 / 126
5G High Capacity Requirements
2. Spectrum Extension (up to 100GHz)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 116 / 126
5G High Capacity Requirements
3. Increase Network DensityMany BSs and many cells of different size
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 117 / 126
5G High Capacity Requirements
4. Overall Performance (Capacity)
Note: with a spatiotemporal array massive MIMO ⇒⇓
Overall >> 10000× Performance
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 118 / 126
5G ITU Recommendations
ITU Recommendations (IMT2020 Spinder Diagram)
Key Capabilities
Peak data rate 20 Gbps
User data rate 100 Mbps
Spectrum Effi ciency 3×Mobility 500 km/h
Latency 1 msConnection density 106 devices
km2
Energy effi ciency 100×Area traffi c capacity 10 Mbits
sm2
Recommendation ITU R M. 2083 0, “IMT Vision Framework and overall objectives of the
future development of IMT for 2020 and beyond,” Sep. 2015.
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 119 / 126
5G ITU Recommendations
ITU Recommendations (Triangle Diagram)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 120 / 126
5G ITU Recommendations
Importance of Key Capabilities
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 121 / 126
5G New Radio
"New Radio"3GPP named 5G as "New Radio " (NR)
NR supports OFDM-family withI subcarrier spacing : 15×2m kHz (Note: In LTE-4G this is 15kHz)
F m = 0 or positive integers: 15, 30, 60, 120, 240, 480 kHzF m are negative integers: 3.75, 7.5 kHz
I Frame length : no frame length (Note: In LTE-4G this is 10ms)
I Subframe length : 1ms (Note: In LTE-4G this is 1ms)
I slot length :F subcarrier spacing ≤ 60 kHz: 7 or 14 OFDM symbols per slotF subcarrier spacing > 60 kHz: 14 OFDM symbols per slotF Note: In LTE-4G this is 7 OFMA symbols
I NR supports emBC, mMC, mCCI NR supports spectrum below and above 6GHz (this could be licensedor unlicenced)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 122 / 126
5G 5G Frame Structure
5G Frame Structure
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 123 / 126
5G Notes
Note-1: Some Current Matlab Useful Toolboxes
Phased Array System Toolbox
Antenna Toolbox
RF Blockset
RF Toolbox
Communications Systems Toolbox
Global Optimisation Toolbox
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 124 / 126
5G Notes
Note-2: Typical Array Design Elements
Array geometry
Array aperture
Lattice structure of the elements and element tapering
array ambiguities
array uncertainties (geometrical and electrical, including mutualcoupling)
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 125 / 126
An Imaginary Illustration
An Imaginary IllustrationThe figure below shows an imaginary illustration of the futurehandheld mobile set with a Monolithic Microwave Integrated Circuit(MMIC) antenna array. This is a 1995 vision of the future.
Prof. A. Manikas (Imperial College) EE401: ST, maMI and mmWave v.19 126 / 126