Halim Yanikomeroglu

1
Infrastructure-Based Wireless Multihop, Relay,
Mesh Networks

• Halim Yanikomeroglu
• halim_at_sce.carleton.ca
• www.sce.carleton.ca/faculty/yanikomeroglu.html

Broadband Communications Wireless Systems
(BCWS) Centre Department of Systems Computer
Engineering Carleton University Ottawa
2
Relaying New Perspective vs Old Perspective
Conventional relaying for coverage Analog/digital
dump (non-selective) repeaters Satellite,
microwave, cellular, deep space,
R
Relay
R
S
R
R
S/R

R

R
Relay/mesh/multihop networks for coverage,
throughput, QoS, , through cooperation


3
Relaying
single-hop
multihop

• Relaying paradigm shift in systems level
• wired AWGN ? wireless fading ? wireless
  mesh
• Impact in all layers of wireless communications
• propagation
• physical layer (PHYITDSP) channel capacity,
  cooperative relaying,
• multiple access layer (MAC) RRM, scheduling,
  CAC,
• networking layer load balacing, routing,
  handoff,
• higher layers and protocols

4
Relaying
single-hop
multihop

• Relaying great interest in academia and industry
• great interest does not necessarily mean
  successful realization
• new network architecture ? network problem
• physical layer just one element
• MAC, networking, protocols important
• time to realization
• TX diversity (Alamouti), turbo codes 4-5 year
• CDMA 15 years
• ad hoc networks 8
• relaying will take some time, but will become
  a reality

5
Relaying Comprehensive Investigation Required

• Simple example
• Relaying links (hops) with less path-loss
• higher spectral efficiency in each link
• Relay cannot receive and transmit at the same
  channel (half-dublex)
• 2-hop link ? 2 channels (time or frequency)
• n-hop link ? 2 to n channels
• Capacity gain or loss?

6
Infrastructure-based vs Infrastructure-less
Multihop Networks
cellular WiFi WiMax (sensor)
ad hoc
infrastructure-based multihop network BS/AP ?
common source or sink
infrastructure-less multihop network
systems networking layers many differences
physical layer many
similarities
7
Outline

• Part I Relaying for cost-effective ubiquitous
  high data rate coverage
• Capacity limited vs coverage limited networks
• Relaying promising solution for coverage limited
  networks
• Part II Possibilities
• Analog relaying vs digital relaying
• Fixed relaying vs terminal relaying
• Homogeneous relaying (single air interface) vs
  heterogeneous relaying (dual air interface)
• Part III Further exploitation of the
  relay/multihop/mesh architecture
• Cooperative relaying
• Novel diversity schemes
• Intelligent routing and scheduling
• Diversity- and AMC (Adaptive Modulation and
  Coding)-aware routing in infrastructure-based
  TDMA multihop networks
• BS-relay coordination
• Dynamic frequency hopping in cellular relay
  networks

8
Expectations for 4G Wireless Networks

• WWRF (World Wireless Research Forum) mITF
  predictions
• mobile up to 100 Mbps
• stationary/nomadic up to 1Gbps !!!

But how?

• More bandwidth is needed
• around 3 or 5 GHz band (World Radio Conference,
  Nov 2007)

• More bandwidth ? more rates it does not scale
  necessarily!
• High bandwidth high carrier frequency
• Tremendous stress on link budget

• advanced antenna technologies (MIMO, smart)
• advanced signal processing (modulation, coding,
  equalization)
• advanced radio resource management techniques

necessary but not sufficient

• A fundamental upgrade in the network architecture
  is needed

9
Cellular Design Fundamentals
QoS
Eb
Pr
I)
R.Eb
R Data Rate
Pathloss
Pt
Cell Size
Propagation Conditions
II)
10
Capacity-Limited Networks
11
Coverage-Limited Networks
available capacity / cell gt capacity demand ?
coverage limited
12
Capacity-Limited vs Coverage-Limited Networks
1G 2G
Capacity limited ? network grows as needed ?
great success

• Ubiquitous high data rate coverage limited
• ? very high deployment cost from the beginning
• ? great challenge

(3G) 4G
WLAN low deployment cost ? great success
13
Coverage Extension through Digital Fixed Relays

• Low cost digital fixed relays located at
  strategic locations
• No wired internet connection at relays
• Different from conventional fixed relays
  (selective relaying)
• Different from ad hoc networks (routing is less
  of an issue)

What is a relay?
relay
relay
relay
relay
BS/AP
BS/AP
Same high data rate coverage
relay
relay
14
Coverage CDFs
H. Hu, H. Yanikomeroglu, D.D. Falconer, S.
Periyalwar Range Extension w/o Capacity Penalty
in Cellular Networks with Fixed Relays,
Globecom 2004
propagation exponent 3
propagation exponent 3
propagation exponent 3.5
15
Average spectral efficiency w.r.t. cell size
Outage w.r.t. cell size
propagation exponent 3.5
16
Cost-Efficient Range Extension

• Same average spectral efficiency and outage
  w.r.t. cell size trends are observed for
  different values of
• Propagation exponent
• Cluster size
• Shadowing standard deviation
• BS transmit power
• Ex Range extension x2 Cost micro-BS / Cost
  relay 10
• Cost microcellular network / Cost relay
  network
• Cost L micro-BSs / (Cost L/4 micro-BSs
  Cost 6L/4 relays)
• 40/16 2.5

Relay networks significant potential for range
extension
17
(IEEE Globecom 2004)
Cost conventional network / Cost relay
network
18
Other Cost Effective Network Architecture
Alternatives
Infostations
Distributed Antennas
Microcellular Network
Antenna Remoting / Radio-on-Fiber
Chinese FuTURE Project
(Future Technology for Universal Radio
Environment) www.chinab3g.org/english/futureproje
ct.htm
non-ubiquitous coverage
wiring cost
wiring cost
Base station
Antenna
19
Capacity of Cellular Fixed Relay/Mesh Networks
Central Node (CN)
?
Relay
Wireless Link

• Only CN is connected to the backhaul

• No Tx Rx on the same channel for a relay

• Nodes have two kinds of antenna

• Direct link with only the neighbor nodes

• Same BW for each primary link

M of root nodes (trees) N of nodes per
tree MN1 of nodes per cell
20
W (Hz) B (Hz) RCCN (bits/sec) RCFRN
(bits/sec) RB M N
Total available bandwidth for a cell
Available bandwidth for a relay
W
B
RCCN RW Capacity of CCN is a function of
W RCFRN (M.N1)RB M.N.RB
21
(No Transcript)
22

23

24

25

26
Total Number Channel Groups
27
Previously it is stated that Then, In
general, if each node has child nodes
28
Total number channel groups when all the hop
links use orthogonal channel groups
Then
29
Capacity Comparisons

• When every other hop links reuse the same
  channel groups

• When all of the links use orthogonal channel
  groups

30
Bandwidth Allocation to Access and Feeder Systems
Florea, Yanikomeroglu IEEE WCNC06
BT Total BW BA total Access BW BF total Feeder
BW Ba access BW for each relay Bf feeder BW for
each relay ma spectral efficiency for access mf
spectral efficiency for feeder nr number of
relays BT BA BF BA NBa BF nr Bf Bf mf
Ba ma
Access system cluster size N 3
Base Station service area
Each color represents a frequency group
Relay service area
31
Bandwidth Allocation to Access and Feeder Systems
Bandwidth allocation for ACCESS
feeder portion
access portion
Bandwidth allocation for FEEDER
if nr ? 8, then BA/BT ? 0 and BF /BT ? 1
Ex nr 25, mf 4, ma 2
Throughput
access, feeder throughput
N 3 19, 81 3.2 BT
N 1 7, 93 3.7 BT
if nr ? 8, then T ? mf BT
32
Bandwidth Allocation to Access and Feeder Systems
Why to use relays?
?
100 Mbps only in this small area
67 Mbps in a much larger area
33
Relay Network vs Microcellular Network
Potential capacity remains more or less the same
Usable capacity increases (outage decreases)
due to better coverage
Relays distribute the total capacity throughout
the coverage region
Relay Network Microcellular Network
Cost-effective high data rate coverage excellent moderate
Capacity (aggregate throughput) moderate excellent
Rule of Thumb in Design Deploy as many
BSs/APs as needed according to capacity demand
Then distribute the capacity in the coverage
region evenly using as many relays as needed
34
Spectral Efficiency of n-hop Link with Orthogonal
Channels
b spectral efficiency of single-hop link ai
spectral efficiency in hop i ac net (overall)
spectral efficiency of n-hop link M message
size B bandwidth T message transfer time
TMH T1 T2 Tn (assuming orthogonal
channels)
TMH M/(Ba1) M/(Ba2) M/(Ban) M/(Bac )
35
Multiplexing Loss in Multihop Relaying with
Orthogonal Channels

• When does it make sense to break a single-hop
  into multiple hops?
• Low SNR case
• Single-hop with 4 dB SNR ? ½-rate QPSK 1 b/s/Hz
• Two hops each with 12 dB SNR ? ¾-rate 16-QAM 3
  b/s/Hz
• Net spectral efficiency 1.5 b/s/Hz ? use
  multihop
• High SNR case
• Single-hop with 26 dB SNR ? full-rate 64-QAM 6
  b/s/Hz
• Two hops each with 34 dB SNR ? 128-QAM 7 b/s/Hz
• Net spectral efficiency 3.5 b/s/Hz ? use
  single-hop
• Rule of thumb low SNR ? multi-hop
• high SNR ? single-hop

opportunistic relaying
36
Multi-Hop Criterion
Florea, Yanikomeroglu IEEE Globecom05

• If single-hop SNR (g) satisfies
• then
• net spectral efficiencyn1 gt
• net spectral efficiencyn
• n - number of hops
• p - path loss exponent
• Assumptions
• all links have same path loss exponent p
• All relays are placed uniformly on a in straight
  line from source to destination

SNR values under which there exists a (n1)-hop
link with better spectral efficiency compared
with an n-hop link
37
Outline

• Part II Possibilities
• Analog relaying relaying vs digital relaying
• Fixed relaying vs terminal relaying
• Homogeneous relaying (single air interface) vs
  heterogeneous relaying (dual air interface)

38
Analog Relaying vs Digital Relaying

• Digital relaying (router, bridge)
• Regenerative relaying
• Decode-and-forward (detect-and-forward) relaying
• Adaptive (selective) decode-and-forward
• Analog relaying
• Non-regenerative relaying
• Amplify-and-forward relaying
• On channel
• With frequency translation
• Analog relaying ? noise propagation
• Digital relaying ? error propagation
• Analog relaying may be better than digital
  relaying in certain scenarios
• Hybrid analog/digital relaying another
  possibility

DF
CMP
digital relay
39
Fixed Relaying vs Terminal/Mobile Relaying

• Terminal relaying rich theoretical area, full of
  potentials
• But, many technical challenges
• No service guarantee
• Increased energy consumption (fast battery
  draining)
• Increased transmit power (in CDMA)
• Additional hardware and functionality (higher
  terminal cost)
• Security issues
• Frequent hand-offs (especially in the presence of
  high mobility)
• Terminal relaying any incentives?
• Special applications single team
  (law-enforcement, military, rescue)
• Non-battery powered fixed user terminals
  (802.16a)
• Cooperative relaying with simultaneous mutual
  benefits (symmetric cooperation)
• Personal area networks
• Commercial applications business plan needed
  (air time offers?)
• Ad hoc networks (infrastructureless) no internet
  connection!

40
Fixed Relaying vs Terminal/Mobile Relaying

• Routing easier in infrastructure-based multihop
  networks than infrastructureless ad hoc multihop
  networks
• Nodes with extra complexity and intelligence
  (BS/AP or fixed relays)
• Common source or destination
• ? Routing with more demanding goals possible
• Expectations in 4G networks
• first, fixed relays
• then, mobile/terminal relays
• Single-hop (infrastructureless) ad hoc networks
  possible
• Multihop (infrastructureless) ad hoc networks
  commercially difficult!

41
Heterogeneous (vs Homogeneous) Relaying

• Decoupling of access and backbone networks
• Access air interface A
• Backbone (feeder) air interface B
• Customized air interfaces
• Easier interference management
• License-exempt bands can be utilized

42
Single Air Interface vs Dual Air Interface
Walsh, Yanikomeroglu IEEE CCECE04
New issues emerge Ex power control in the
reverse-link of two-hop CDMA networks
Two-hop PC
Single-hop PC
43
Outline

• Part III Further exploitation of the
  relay/multihop/mesh architecture
• Novel diversity schemes
• Advanced cooperative relaying
• Intelligent routing
• Diversity- and AMC (Adaptive Modulation and
  Coding)-aware routing in infrastructure-based
  TDMA multihop networks
• BS-Relay coordination
• Dynamic frequency hopping in cellular relay
  networks
• WiMax mesh networks

44
Theoretical Literature on Relay/Mesh/Multihop
Networks

• Van der Meulen (68,71)
• El Gamal, Cover, Aref (79,80,82) Stanford
• Willems (83,85)
• Sendonaris, Erkip (98-) Rice
• Laneman Notre Dame, Wornell MIT (00-)
• Gupta ASU, Kumar UIUC
• Dohler KCL
• Nosratinia, Hunter UTD
• Tse Berkeley
• Hasna Qatar, Alouini Minnesota
• Giannakis, Cai, Ribeiro Minnesota
• Host-Madsen Hawaii
• Pottie
• Zhang ASU
• Stefano Brooklyn
• Wittneben, Rankov
• Cho, Haas Cornell

• Anghel, Emamian, Kaveh Minnesota
• Bolcskei, Nabar ETH
• Gastpar Berkeley
• Dawy AUB
• Kramer Bell Labs
• Franceschetti UCSD
• Herhold, Zimmermann TUD
• Vetterli EPFL
• Valenti
• Walke Aachen
• Karagiannidis AU Thessaloniki

A few conference papers ? explosion in literature
45
Relaying New Perspective vs Old Perspective
coverage extension
R
Relay
R
S
R
R
S/R

R

R
Cooperative relaying ? virtual antenna
array diversity, space-time coding, MIMO,


46
Asymmetric vs Symmetric Cooperation

• Asymmetric Cooperation relay terminal between
  source and destination
• Only one terminal benefits
• Pathloss gain
• Symmetric Cooperation a pair of nearby terminals
  cooperate
• Immediate benefit for both terminals
• No pathloss gain

47
User Cooperative Diversity
Sendonaris, Erkip, Aazhang IEEE T-COM, Nov.
2003 (ISIT 1998)
Channel Model (full-duplex assumption)
48
Implementation case example CDMA Resources
Distribution codes
power allocation and
No cooperation

• Using the same number of codes as in no
  cooperation scenario

With cooperation
Users exchange cooperative information BS may
hear
Constructed cooperative signals are sent to BS
Users transmit to BS
49
Laneman, Tse, Wornell IEEE T-IT,
Dec04 Laneman, Wornell IEEE T-IT, Oct03
50
Parallel Relays
relay
first slot broadcast
second slot multiple access orthogonal ?
multiplexing loss beamforming ? CSI (overhead)
relay
diversity order up to R1 provided that error
propagation is prevented
relay

51
Multi-Antenna Aspects of Cooperative Fixed Relays
Adinoyi, Yanikomeroglu WCNC06 (IEEE
TWireless07)
use few relays with multi-antennas (even with
selection combining) instead of many relays with
single antennas
Rlay 1
First Hop
Second Hop

52
Practical Cooperative Communication Schemes
Through Fixed Relays
Adinoyi, Yanikomeroglu WWRF15, VTCS06
no need for incentives for cooperation no
security risk no complexity incurred in
terminals complexity moved to network
53
Practical Cooperative Communication Schemes
Through Fixed Relays
54
Practical Cooperative Communication Schemes
Through Fixed Relays
55
Multihop Diversity

• Analysis of multihop channels with diversity
• Decoded relaying with diversity intermediate
  terminals combine, digitally decode and re-encode
  the received signal from all preceding terminals
• Amplified relaying with diversity intermediate
  terminals combine and amplify the received signal
  from all preceding terminals
• Diversity in 2-Hop Links
• Mazen Hasna, Mohamed-Slim Alouini
• J. Laneman, Greg Wornell
• Diversity in n-Hop Links (for fully connected
  networks)
• John Boyer, David D. Falconer, Halim
  Yanikomeroglu, Multihop Diversity in Wireless
  Relaying Channels, IEEE Trans. on
  Communications, Oct. 2004
• Main Observations
• Comparison of relaying and diversity schemes wrt
  BER
• Multihop Diversity gt Multihop gt Singlehop
• Amplified Relaying gt Decoded Relaying
• DRMD improves when intermediate terminals are
  closer to the source terminal
• ARMD improves when intermediate terminals are
  closer to the destination terminal

56
Full-Diversity Relays and Destination
Boyer, Falconer, Yanikomeroglu IEEE T-COM, Oct.
2004

• Full diversity reception at all receivers
• Requires n channels for n relaying hops
• Complex relay behavior (diversity combining)

57
Aggregate SNR of Amplified Relaying Channels

• John Boyer, David D. Falconer, Halim
  Yanikomeroglu, On the Aggregate SNR of
• Amplified Relaying Channels, IEEE Globecom 2004.
• Aggregate Signal to Noise Ratio
• Aggregate inclusion of propagated noise terms
  in the SNR formulation.
• Propagated noise terms are generated as amplified
  relaying terminals amplify both the information
  and noise portions of received signals
  indiscriminately.
• Motivated by findings indicating that the
  performance of amplified relaying can approach
  and in some cases exceed that of decoded
  relaying.
• Aggregate SNR expressions developed for amplified
  relaying channels with given source, destination,
  and relaying terminals, link connectivity, link
  attenuation, transmit power, and receiver noise.
• Aggregate SNR expression developed for following
  network connectivity
• Serial Amplified Relaying Channels (serial node
  connectivity)
• Parallel Amplified Relaying Channels (parallel
  node connectivity)
• General Amplified Relaying Channels (general node
  connectivity)

58
Amplified Relaying
Source terminal
Destination terminal
59
ith Amplifying Relay
From Terminal k
60
Definitions

• Aggregate SNR at for Link SNR at
  for
• where
• is the transmitted power
• is the complex amplitude of the information
  symbol
• is propagated noise
• is distance-dependant attenuation,
  shadowing, and fading
• is the variance of the zero-mean Gaussian
  random variable
• This model normalizes the transmit signal such
  that

61
Some Equations

• Serial Amplified Relaying Channels (recursive
  form)
• Serial Amplified Relaying Channels (sum of
  products form)
• Parallel Amplified Relaying Channels
• General Amplified Relaying Channels

62
Example Channels
Fig 1. Serial Connected
Fig 2. Parallel Connected (multi-route diversity)
Fig 3. General Connected
Fig 4. General Connected 2
Fig 5. Fully Connected
63
Aggregate SNR vs. Link SNR
Fig. 1 Fig. 2 Fig. 3 Fig. 4 Fig. 5
64
Comparison with a Linear Relation (high order
multiplicative termsin formula at bottom of
slide 62 removed)
Fig. 1 Fig. 2 Fig. 5
65
Observations

• Serial Amplified Relaying Channels Allocate
  relatively more power to weaker links
• Parallel Amplified Relaying Channels Allocate
  relatively less power to weaker links
• General Amplified Relaying Channels Allocate
  relatively more power to weak links that are not
  parallel to strong links and relatively less
  power to weak links that are parallel to strong
  links
• Maximizing Links in parallel (diversity gains)
  more important than maximizing links in serial
  (attenuation gains) for the example simulations.
• For strong links in serial the performance is
  approximately linear with respect to the
  component link SNRs
• For weak links in serial the performance is less
  than linear with respect to the component link
  SNRs
• Amplified relaying may not be appropriate for
  very low SNR systems as propagated noise becomes
  limiting factor.

66
System Resource Constraints

• John Boyer, David D. Falconer, and Halim
  Yanikomeroglu, Impact of System Resource
  Constraints on the Connectivity of Wireless
  Relaying Networks, ICC 2005 (under review in
  TWireless)
• System resource constraints have a direct impact
  on the link connectivity of wireless relaying
  networks and therefore the possible distributed
  spatial diversity techniques
• Constraints considered
• Number of Channels Available (NCA) N or 2
  available channels
• Relay Combination (RC) Diversity combining or
  simple relay
• Destination Combination (DC) Diversity combining
  or simple relay
• Multiple Channel Reception (MCR) Multiple
  channels or single combined by receiver
• Multiple Channel Transmission (MCT) Multiple
  channels or single transmitted
• Possible constraint combinations are analyzed and
  reduced
• Relaying types
• Amplify-and-forward (analog) relaying
• Decode-and-forward (digital) relaying with error
  propagation
• Decode-and-forward (digital) relaying without
  error propagation

67
Full-Diversity Relays and Destination
Boyer, Falconer, Yanikomeroglu IEEE T-COM, Oct.
2004

• Full diversity reception at all receivers
• Requires n channels for n relaying hops
• Complex relay behavior (diversity combining)

68
2-Channel Diversity Relays and Destination

• Successive transmitters along multihop path
  transmit/receive on alternating channels.
• Requries 2 channels for n relaying hops.
• Complex relay behavior (diversity combining).

69
2-Channel Diversity Relays w Full Diversity
Destination

• Alternating channels with full diversity at
  destination
• Requires 2 channels for n relaying hops
• Complex relay behavior (diversity combining)

70
Non-Diversity Relays w Full Diversity Destination

• Alternating channels with full diversity at dest.
• Requries 2 channels for n relaying hops.
• Simple relay behavior (no combining).

71
System Connectivity Models

• Models defined by the connectivity of source,
  relay, destination
• 1X Terminal class X connected to one transmitter
• 2X Terminal class X connected to the subset of
  transmitters on one channel
• FX Terminal class X connected to all
  transmitters
• Inter-model transitions based on minimum cost
  constraint sets

S Source R Relay D Destination
72
System Connectivity Models (Samples)
1
1
0
1
0
1
0
1
1
0
0
1
1
0
0
0
1R1D
2R2D
0
1
1
1
0
1
0
0
1
0
0
1
1
1
1
0
0
0
1
0
0
0
1
1R2D
2RFD
0
1
1
0
0
0
1
1
0
0
1
0
1RFD
FRFD
73
Minimum Cost Constraint Sets

• Mark of X indicates that the constraint is
  lifted ? the capability described by the
  corresponding system resource is available

Model NCA N Channels Available RC Relay Combination DC Destination Combination MCR Multiple Channel Reception MCT Multiple Channel Transmission
1R1D
1R2D X
1R2D2S X X
1RFD X X
2R1D X
2R1D2S X X
2R2D X X
2R2DFS X X X
2RFD X X X
2RFDFS X X X X
FR1D X X X
FRFD X X X X
74
Amplified Relaying
1R1D
Singlehop
1R1D
1R2D
FR1D
2R2D
2R2D
1RFD
1RFD
FRFD
FRFD
2RFD
2RFD
75
Decoded Relaying with Error Propagation
1R1D
1R2D
1RFD
Singlehop
2R2D
2RFD
FRFD
1R1D
FRFD
76
Decoded Relaying without Error Propagation
1R1D
Singlehop
1R2D
1R1D
1RFD
2R2D
2R2D
FRFD
FRFD
2RFD
77
Observation I

• Simulation of connectivity models for amplified
  relaying, decoded relaying with error
  propagation, and decoded relaying without error
  propagation isolates the impact of each
  constraint
• Impact summary of system resource constraints
• Relaying Method NCA RC DC MCR MCT
• Amplified Small Small Large Large Small
• Decoded w Prop Small Medium Medium Small Medium
• Decoded w/o Prop Small Medium Large Large Small
• Different impact for different relaying methods.
• Connectivity priorities for optimal performance
• Amplified ? Maximize destination connectivity
• Decoded w Prop ? Equalize relay destination
  connectivity
• Decoded w/o Prop ? Maximize destination
  connectivity disjoint network paths

78
Observations II

• Diversity order of the system is dependent on the
  connectivity
• Amplified relaying ? Dependent only on
  destination
• Decoded relaying w Prop ? Dependent only on
  minimally connected relay
• Decoded relaying w/o Prop ? Dependent on
  destination and disjoint paths
• Guidance for the order in which the constraints
  should be lifted
• Amplified ? DC, MCR, RC, MCT
• Decoded w Prop ? RC, DC, MCT, MCR
• Decoded w/o Prop ? DC, MCR, RC, MCT
• Impact of lifting NCA contraint when all other
  constraints are lifted is small
• Likely will not be implemented in practice since
  the cost of N orthogonal channels per information
  signal transmission is very high with respect to
  spectral efficiency.

79
Ongoing Work

• Incorporation of Interhop Interference
  Cancellation (IC) as a constraint
• Explicit separation of Common Channel Combination
  and Orthogonal Channel Combination as constraints
• Generalization of analysis for arbitrary number
  of channels available (KltN)
• Formalization of method for deriving system
  connectivity models
• Mapping of models to cooperative diversity
  techniques in the literature
• Interrelationships between system resource
  constriants
• Relationship of system resource constraints to
  relay node placement
• Grid placement
• Colinear equidistant between source and
  destination placement
• Bunched nodes placement between source and
  destination
• Random placement with varying uniformity of relay
  density
• Practical issues and cost of system resource
  constraints
• Common channel combination techniques
• Cooperative coding
• Relay block size
• Spatial reuse of channels
• Wireless fading models
• Spectral efficiency

80
Coordination among BSs/APs and Relays

• Coordination among BSs
• Scheduling
• Interference management
• Radio resource management
• Admission control
• Rich literature
• Limited usage in practice in conventional
  cellular networks
• May be used in cellular relay networks

81
CLASSICAL DYNAMIC FREQUENCY HOPPING WITH NETWORK
ASSISTED RESOURCE ALLOCATION (DFH with NARA)
ATT Bell Labs
i.
ii.
Each terminal measures path losses to the
neighboring bases and transmits this information
to its serving base on a regular basis.
Each base communicates to several tiers of its
neighbors the information about its own resource
utilization (i.e. time slots, frequency hopping
patterns and current power levels).
iii.
The serving base station calculates the
interference level at each available resource,
determines the least-interfered time slot and FH
pattern pair, and assigns this to the terminal.
82
BLOCK DIAGRAM OF A CELLULAR SYSTEM THAT SUPPORTS
DFH WITH NARA FOR DOWNLINK
MOBILE STATION 1
.
Measure Pathloss on BCCH for BS 1
Measure Pathloss on BCCH for BS K
.
Read and Use Specified FH Pattern
Average
Average
.
Send to BS1
Send to BS1
WIRELESS MEDIUM
To Other Mobiles
From Other Mobiles
BASE STATION 1
Send Orders To MSs with Next FH Patterns
Collect Measurements From All MS in this BS
Coverage Area
Local Copy of Measurements For All MS from All BS
Local Copy of FH Patterns For All BS
Manage Frequency Hop Patterns For This BS
From Other Base Stations
From Other Base Stations
To Other Base Stations
To Other Base Stations
LANDLINE NETWORK
83
(No Transcript)
84
DFH with LIMITED INFORMATION (Time Slot 2)
Mubarek, Yanikomeroglu, Periyalwar

• R1 3 in-cell interferers (R2,R6,BS) and 3
  out-of-cell interferers (R3,R4,R5)
• UE pathloss info R1?BS
• BS already has resource utilization information
  of the in-cell interferers of R1
• BS decide on DFH pattern based on limited info
• BS?R1 DFH pattern

85
FIGURE 9
INTERFERING SUB-CELLS
First Tier Interferers
Second Tier Interferers
Base Station
Relay Station
User Equipment
Cell Border
Relay or Base Neighborhood
86
UE
Measure Pathloss for BS
Measure Pathloss for Interfering Relay1
Measure Pathloss for Interfering Relay2
Get the new FH Pattern
UE in RS-Neighborhoods
Average out Rayleigh
Average out Rayleigh
Average out Rayleigh
BASE STATION
Notify RSs with the new FH Patterns for UEs in RS
Neighborhoods
UEs IN RS NEIGHBORHOODS
Collect Pathloss Reports from UEs in RS
Neighborhood
Create new Random FH Patterns for UEs in RS
Neighborhoods from the pool of available and
unblocked Resources
Calculate which resources result in a SIR less
than the threshold SIR, SIRTh, and block them
NO COORDINATION
Notify UEs in RS Neighborhoods with their new FH
Patterns
IN-CELL RELAYS
LANDLINE NETWORK
Get the new FH Pattern
OTHER BASE STATIONS
OUT-OF-CELL RELAYS
87
DFH with LIMITED INFORMATION (Time Slot 2)
UE in BS service region DFH with full information
88
UE
Measure Pathloss for RS1

Measure Pathloss for RS6
BS-Neighborhood, BS relays as interferers
Get the new FH Pattern

Average out Rayleigh
Average out Rayleigh
BASE STATION
Notify UEs in BS Neighborhood with their new FH
Patterns
Collect Pathloss Reports from UEs in BS
Neighborhood
UEs IN BS NEIGHBORHOOD
Create new FH Patterns for UEs in BS Neighborhood
NO COORDINATION
LANDLINE NETWORK
IN-CELL RELAYS
OTHER BASE STATIONS
OUT-OF-CELL RELAYS
89
Cooperative Induced Multi-user Diversity Relaying
(CIMDR)
Navaie, Yanikomeroglu VTCS06

• First hop
• Multi-user diversity exploits through
  transmission with maximum bit-rate
• Second-hop
• Multi-user diversity exploits through
  transmission on a good channel
• Two phase protocol Feeding and delivery phase

90
Performance Improvements through the Mesh
Architecture in TDMA based Broadband Fixed
Cellular Network
Syed, Ahmed, Yanikomeroglu, Mahmoud WCNC04
91
Cellular Mesh Network with Global Resource
Allocation
92
Algorithm for Constructing Routing Table
Step 1 Reject all node-node node-BS links with
PLgtPLmax
Step 2 List all 2-hop 3-hop routes between
source node and BS(s)
Step 3 Arrange in ascending order,the routes
found in Step 2 using criterion min max (PLi )
where i1,2 or i1,2,3
Step 4 If tie in max(PLi), then min ? PLi
route on the top
93
Route Selection

• Policy
• Minimum Number of Hops Route First
• Conditions
• Free slot(s) available on all hops
• SINRr?SINRth on the free slot(s)
• Salient Features
• Less spectral resources used (time slots)
• Viable SINR links used (SINRr?SINRth)
• Simple call admission policy
• Simple algorithm

94
Adaptive Modulation Coding
SINR (dB) Code Mod. Info. bps/hz
lt4.65 - 0
4.65-7.45 ½ QPSK 1
7.45-10.93 ¾ QPSK 1.5
10.93-12.0 ½ 16-QAM 2.0
12.0-14.02 2/3 16-QAM 2.67
14.02-15.0 ¾ 16-QAM 3
15.0-17.7 7/8 16-QAM 3.5
17.70-19.0 2/3 64-QAM 4
19.0-21.94 ¾ 16-QAM 4.5
21.94-26.0 7/8 64-QAM 5.25
gt26.0 64-QAM 6
95
Main Simulation Parameters
System Parameter Simulation Value
Network Cellular, Noise limited, 200 Nodes, 4-Square Cells, Cell Size 3x3 km2
Multiple Access TDMA / FDD, 10 slots/channel
Propagation Channel nn-BS 3.8, nn-n 4, ?n-BS 6dB, ?n-n 4dB
Carriers Bandwidth Single Carrier _at_ 2.5GHz, BW 5 MHzs No. of Channels 2 to 6
Antenna Type 30o, switched beam, Gml 7 dB, Gsl,bl 0 dB, Rooftop (node end)
Transmit Power Fixed, Pt 2 watts
Noise AWGN, Noise Power - 130 dBW
96
Simulation Parameters Cont.
System Parameter Simulation Value
Network Traffic Traffic Arrival Poisson, ?400-8000 burst/sec Burst size Exponential, ?15 kbits
PLmax for Routing 126 dB
SINRth 4.65 dB
Frame specification Tf10ms, 10 slots/frame
Slot allocated per hop 1
Upper limit on consecutive frame drop before retransmission 3
97
Simulation Assumptions

• Snap shot processing at frame level
• All transmissions are slot synchronized
• Independent fixed shadowing on all links
• Doppler shift negligible
• Multipath fading handled by micro diversity
• Infinite buffer size on the node
• All user nodes are active
• Separate control channels are available
• Continuous ARQ Protocol

98
Outage Probability
99
Outage Analysis of SH MH Network, 2-Channels
100
Outage Analysis of SH MH Network, 4-Channels
101
Outage Analysis of SH MH Network, 6-Channels
102
Connectivity Analysis MH Network, 2-Channels
103
Connectivity Analysis MH Network, 4-Channels
104
Connectivity Analysis MH Network, 6-Channels
105
Net Node Throughput
106
Diversity- and AMC (Adaptive Modulation and
Coding)-Aware Routing inInfrastructure-based
TDMA Multihop Networks
Hares, Yanikomeroglu, Hashem VTCF03 Globecom03
4
1
3
2
6
0 AP
5
Time Domain MAC Frame
Connection to 3
Connection to 4
Connection to 6
Hop 0 to 4
Hop 0 to 1
Hop 1 to 2
Hop 2 to 3
Hop 0 to 5
Hop 5 to 6
Extra channels are not used. Connections and hops
are orthogonal in the time domain.
107
Routing

• Routing objective
• Select relay nodes and hop modulation/coding (MC)
  to maximize throughput
• Throughput (Information Rate bits/sec)(1 -
  Probability of error)
• Increase Info. Rate or decrease end-to-end error
  rate to increase throughput. How?

108
Frame Allocation for Relaying

• Adaptive modulation and coding (AMC)
• Different MC used on hops
• Amount of data entering and exiting relaying
  nodes are equal

QAM4
Hop0
Hop0
Hop0
2
0
QAM64
QAM64
Hop0
Hop1
Hop0
Hop1
Hop0
Hop1
0
1
2
109
Multihop Diversity
Time Domain
QAM64
QAM16
QAM64
Hop1
Hop2
Hop0
Hop2
Hop1
Hop1
Hop0
Hop0
0
1
2
3
Node 2 Receiver Operation Equivalent
Symbols
Yes
Bits
Hop0
Decoder
CRC OK?
No
Symbols
Yes
Bits
Hop1
Decoder
CRC OK?
Node 3 Receiver Operation Equivalent
Hop0
MRC
Symbols
Yes
Bits
Decoder
CRC OK?
Hop2
No
Symbols
Yes
Bits
Hop1
Decoder
CRC OK?
110
Adaptive Modulation Coding Maximization (AMCM)

• Originally, hop modes were selected to maximize
  data rates on hops.
• AMCM adapts hop modes to maximize the connection
  throughput for systems using MRC diversity.
• Performed after route has been selected.
• Possible modes a hop can assume is limited to the
  set of modes used in the connection (i.e. QAM16,
  QAM64, QPSK1/2).
• For each iteration, examine all possible modes
  for all hops.
• Change a mode for a single hop that generates the
  maximum metric for the connection.
• Use the new set of modes for the subsequent
  iteration.
• Stop when a mode change does not increase the
  metric for the connection.

Hop1
Hop2
Hop3
QAM64
QAM16
QPSK1/2
From all possible changes, changing the mode of
hop 2 from QAM64 to QAM16 generates the max.
metric.
Hop1
Hop3
QPSK1/2
QAM16
Use the new set of modes to continue to maximize
the metric.
Stop when mode changes do not increase the
connection metric.
111
Routing Types

• Route Selection Strategies
• Single Hop (SH)
• Multihop (MH)
• Multihop Selection Combining Diversity (MHSC)
• Routing metric factors selection combining
  diversity
• Multihop MRC Diversity (MHMRC)
• Routing metric factors MRC diversity
• Multihop Adaptive Modulation MRC Diversity
  (MHAMMRC)
• Routing metric factors MRC diversity
• Uses AMCM
• Hybrid Digital and Analog Relaying (HDAR)
• Nodes relay incorrectly decoded signals as analog
  signals

112
Routing Example (1)
Iteration
Next 1
Current 0

• Initially, routes only contain the AP and
  destination node.
• Examine all next routes.
• Next routes (black) built off current routes
  (red).
• Next routes generating max. metrics are in
  purple.
• If metric of next route gt metric of current
  route, on next iteration, current route next
  route.
• Next iteration routes
• B AP-A-B
• C AP-E-C
• E AP-A-E

AP
A
A
B
B
C
C
D
D
E
E
113
Routing Example (2)
Iteration
Next 2
Current 1

• Only check routes to nodes not included in
  current route.
• Next iteration routes
• C AP-A-E-C
• D AP-E-C-D
• Need to only examine next routes built from
  routes which changed on previous iteration.

AP
A
A
B
B
C
C
D
D
E
E
E
114
Routing Example (3)
Iteration
Next 3
Current 2

• Stop searching when next generation of routes do
  not yield higher metrics.
• Final connections
• A AP-A
• B AP-A-B
• C AP-A-E-C
• D AP-E-C-D
• E AP-A-E

AP
A
A
B
B
C
C
D
D
E
E
115
Simulation - Routing Example
116
Simulation - Routing Example
117
Simulation - Parameters

• NLOS office environment
• ETSI-A channel model
• Rayleigh fading 50ns RMS delay spread
• Noise power, PN0 -90dBm
• Propagation exponent, a 3.4
• Carrier frequency, fc 5.3GHz
• Shadowing, s 5.1dB
• Omni-directional antennas
• Fixed transmit power, Ptx 23dBm
• Adaptive modulation
• Constant interference
• Hexagonal radius, R 128m
• Cluster size, N 12
• No mobility

118
Simulation Throughput, 128m Cell
119
Simulation - Number of hops, 128m Cell
120
Simulation Throughput, 256m Cell
121
Simulation - Number of hops, 256m Cell
122
Simulation Results
Routing Type System Diversity Avg. Network Throughput Mbps Avg. Network Throughput Mbps Avg. Hops in Route Avg. Hops in Route
Routing Type System Diversity 128m Cell 256m Cell 128m Cell 256m Cell
SH None 7.75 2.07 1 1
MH None 12.77 4.17 2.21 2.93
MHSC SC 13.17 4.70 2.64 4.17
MHMRC MRC 13.19 4.70 2.62 4.14
MHAMMRC MRC 13.26 4.85 2.62 4.24
MHAMMRC-HDAR MRC 14.32 5.62 2.56 3.70

• Routing Type routing/metric type
• System Diversity form of diversity used at
  nodes
• SH singlehop, MH multihop (routing algorithm)
• MHAM multihop adaptive modulation (routing
  algorithm)

123
Observations

• Multihop routes ? Optimal 2-hop routes
• Diversity techniques very attractive -- they do
  not use additional radio resources (power or
  bandwidth)
• Routing incorporates diversity benefits
• HDAR increases diversity benefits
• Average aggregate throughput increases 2-3X
• Outage reduces very significantly (range
  extension remarkable)
• Strategically placed fixed relayers may be very
  attractive
• Can be used in any TDMA network, if
• PER models are known channel updates supported
• For more information
• Shoaev Hares, Halim Yanikomeroglu, and Bassam
  Hashem, "Diversity- and AMC (Adaptive Modulation
  and Coding)-Aware Routing in TDMA Multihop
  Networks", IEEE GLOBECOM 2003
• Shoaev Hares, Halim Yanikomeroglu, and Bassam
  Hashem, "A relaying algorithm for multihop TDMA
  TDD networks using diversity", IEEE VTC Fall 2003

124
Previous Standardization Efforts

• Opportunity-driven multiple access (ODMA) 1999,
  3GPP
• HiperLAN/2 (non-contention based multiple access)

125
Current Interest in Relay/Mesh/Multihop Networks
(1)

• IEEE 802.11s WLAN (Wireless Local Area Network)
  ESS Mesh Networking
• Auto-configuring multihop paths between APs in a
  wireless distribution system.
• Targeted to be approved by 2008.
• IEEE 802.15.5 WPAN (Wireless Personal Area
  Network) Mesh Networking
• Aims at determining the necessary mechanisms that
  must be present in the PHY and MAC layers of
  WPANs to enable mesh networking.
• Targeted to be approved by 2007.
• IEEE 802.16 WMAN (Wireless Metropolitan Area
  Network)
• 802.16-2004 standard Air Interface for Fixed
  Broadband Wireless Access Systems approved in
  July 2004. MAC layer supports an optional mesh
  topology.
• 802.16e amends 802.16 to support mobility for the
  devices operating in the 2-6 GHz licensed bands.
• An optional mesh mode is being considered based
  on 802.16e-2005 OFDMA
• MMR-SG Mobile Multihop Relay Study Group ?
  802.16j
• (Taipei, Sep05 Vancouver, Nov05 New Delhi,
  Jan06 Denver, Mar06)
• http//grouper.ieee.org/groups/802/16/sg/mmr/
• http//ieee802.org/16/sg/mmr/
• IEEE 802.20 MBWA (Mobile Broadband Wireless
  Access)
• Aims at developing the specification of PHY and
  MAC layers of an air interface for interoperable
  mobile broadband wireless access systems,
  operating in licensed bands below 3.5 GHz,
  optimized for IP-data transport, with peak data
  rates per user in excess of 1 Mbps.

http//grouper.ieee.org/groups/802/11/index.html h
ttp//www.802wirelessworld.com
126
Current Interest in Relay/Mesh/Multihop Networks
(2)

• Cellular 4G Networks (WINNER Project)
• Propriety solutions by industry
• BelAir, Firetide, Strix, Tropos, RoamAD, Mesh
  Networks,
• Nortel, Nokia, IBM,
• Booming literature

127
WINNER Wireless World Initiative New Radio

• Integrated Project funded by European Union under
  the 6th Framework Program (FP6)
• Objective to develop a ubiquitous radio system
  concept based on global requirements for mobile
  communication systems beyond 3G. The project
  covers a full scope from short-range to wide-area
  scenarios and will provide significant
  improvement to current systems in terms of
  performance, efficiency, coverage and
  flexibility.
• 01 Jan 2004 31 Dec 2009 (three 24-month phases)
• 50 partners (all European, except 2 Chinese 1
  Canadian Carleton)
• Manufacturers, network operators, academic
  institutions and research centres
• Including Siemens, Alcatel, DoCoMo Europe,
  Ericsson, Nokia, France Telecom, Fujitsu Europe,
  IBM Europe, Philips, Samsung Europe, Vodafone,
  Qualcom Europe, Nortel Europe
• Relaying integral part of WINNER network
  deployment concept
• Check https//www.ist-winner.org ? Public
  Deliverables ? D3.x

128
Concluding Remarks

• Infrastructure-based multihop networks
  cost-effective ubiquitous high data rate coverage
  in future wireless networks
• Fixed relay stations with add-on terminal relays
• Impact in all layers of wireless communications
• Propagation, PHY, MAC, networking, higher layers
  and protocols
• Relay networks will soon become a reality
• Goal to develop advanced cooperation protocols
  and algorithms among relays and APs to obtain
  further performance gains at
• physical layer (cooperative diversity, virtual
  antenna arrays, )
• systems layer (interference avoidance and
  management)
• networking layer (smart scheduling and routing,
  load balancing, )
• by relying on other advanced technologies, such
  as OFDM(A) and MIMO, as much as possible.

129
Tutorial/Overview/Perspective Papers

• H. Yanikomeroglu "Fixed and mobile relaying
  technologies for cellular networks", Second
  Workshop on Applications and Services in Wireless
  Networks (ASWN'02), pp. 75-81, 3-5 July 2002,
  Paris, France.
• H. Yanikomeroglu, "Cellular multihop
  communications infrastructure-based relay
  network architecture for 4G wireless systems",
  the 22nd Queen's Biennial Symposium on
  Communications (QBSC'04), 1-3 June 2004, Queen's
  University, Kingston, Ontario, Canada invited
  paper.
• R. Pabst, B. H. Walke, D. C. Schultz, P. Herhold,
  H. Yanikomeroglu, S. Mukherjee, H. Viswanathan,
  M. Lott, W. Zirwas, M. Dohler, H. Aghvami, D. D.
  Falconer, and G. P. Fettweis, Relay-based
  deployment concepts for wireless and mobile
  broadband radio, IEEE Communications Magazine,
  vol. 42, no. 9, pp. 80-89, September 2004.
• R. Bruno, M. Conti, and E. Gregori, Mesh
  networks commodity multihop ad hoc networks,
  IEEE Communications Magazine, vol. 43, vol. 3,
  pp. 123-131, March 2005.

130
Relay/multihop/mesh networks research at Carleton
University

• Halim Yanikomeroglu
• 2 PDF
• 7 Ph.D. students
• 2 M.A.Sc. students
• 6 B.Eng. students
• visiting professors and researchers
• halim_at_sce.carleton.ca
• www.sce.carleton.ca/faculty/yanikomeroglu.html