• No results found

The dynamic power consumption data of the LDPC decoder refers to the nominal case of Vdd =1.2 V and maximum clock frequency 299 MHz. The input variables are:

• Sub-block size Z; • Code rate R. Z =27 R 1/2 2/3 3/4 5/6 79.27 77.53 76.70 74.55

Table A.4: LDPC decoder power consumption in mW (Z=27).

Z =54 R

1/2 2/3 3/4 5/6

112.65 108.42 107.22 104.53

Table A.5: LDPC decoder power consumption in mW (Z=54).

Z =81 R

1/2 2/3 3/4 5/6

137.52 134.35 134.87 121.61

Acronyms

3G third generation 4G fourth generation

I/O input/output

AGU address generation unit

AMC adaptive modulation and coding ARQ automatic repeat-request

ASIC application-specific integrated circuit BCJR Bahl, Cocke, Jelinek, Raviv

BICM bit-interleaved coded modulation

BICM-ID bit-interleaved coded modulation with iterative decoding BLER block error rate

CDMA code division multiple access

CMOS complementary metal oxide semiconductor CN check node (in a Tanner graph)

CPU central processing unit CRC cyclic-redundancy check CWER codeword error rate

DIBL drain-induced barrier lowering DRC design-rule check

DSP digital signal processing

DVB-H Digital Video Broadcasting - Handheld

DVB-S2 Digital Video Broadcasting - Satellite - Second Generation DVS dynamic supply voltage scaling

ECC error-correcting code

EDGE Enhanced Data Rates for GSM Evolution EP expectation propagation

ETWS early-termination window size

FCC Federal Communications Commission FEC forward error correction (or correcting)

FER frame error rate

FPGA field-programmable gate array FSD fixed-complexity sphere decoding FSM finite state machine

GALS globally asynchronous locally synchronous GDSII Graphic Database System II

GMSK Gaussian minimum shift keying GPRS General Packet Radio Service GPS Global Positioning System

GSM Global System for Mobile Communications HDL hardware description language

HSDPA High-Speed Downlink Packet Access HSPA+ Evolved High-Speed Packet Access i.i.d. independent and identically distributed IC integrated circuit

IDD iterative detection and decoding

IEEE Institute of Electrical and Electronics Engineers LAN local area network

LDD layered detection and decoding LDPC low-density parity check

LL low leakage

LLR log-likelihood ratio LTE Long Term Evolution LVS layout vs. schematic MAC media access control MAP maximum a posteriori MCMC Markov chain Monte Carlo MCS modulation and coding scheme

MF maximum first

MIMO multiple input multiple output

ML maximum likelihood

MMSE minimum mean square error

MP message passing

MPW multi-project wafer NCU node computation unit

OFDM orthogonal frequency division multiplexing OFDMA orthogonal frequency division multiple access OMS offset min-sum

ONPC one node per cycle PAN personal area network

PC personal computer

PCCC parallel concatenated convolutional code PE processing element

PIC parallel interference cancellation QAM quadrature amplitude modulation

QC quasi cyclic

QoS quality of service

QRD QR decomposition

RF radio frequency

RTL register transfer level

SCCC serial concatenated convolutional code SD sphere decoder (or decoding)

SDR software-defined radio

SE Schnorr-Euchner

SNR signal-to-noise ratio SoC system on a chip SP standard performance SPA sum-product algorithm SRAM static random-access memory STS single tree search

UMTS Universal Mobile Telecommunications System

US United States

VHDL very high speed integrated circuit hardware description language VLSI very large scale integration

VN variable node (in a Tanner graph)

WiMAX Worldwide Interoperability for Microwave Access WLAN wireless local area network

Notation – Signal Processing

d diversity gain

MT number of transmit antennae MR number of receive antennae

O set of all the complex scalar symbols belonging to a given constellation

OMT set of all the complex symbol vectors originating from a given constel- lation

Q number of coded bits per complex modulated scalar symbol Es normalised average energy per complex scalar transmit symbol No power spectral density of the additive white Gaussian noise

b information bitstream, input to the encoder on the transmitter side

c coded bitstream, output by the encoder on the transmitter side

s transmit symbol vector

H MIMO channel matrix

n additive noise vector

y received symbol vector

yi complex received symbol on receive antenna i ˆ

s estimated transmit symbol vector, output by the detector on the receiver side

ˆb estimated information bitstream, output by the decoder on the receiver side

Iidd total number of IDD iterations, defined by the number of detector/ decoder runs

iidd index of the current IDD iteration

λa vector of a priori LLRs

λa,det vector of a priori LLRs input to the detector λa,dec vector of a priori LLRs input to the decoder λp vector of a posteriori LLRs

λp,det vector of a posteriori LLRs computed by the detector λp,dec vector of a posteriori LLRs computed by the decoder λe vector of extrinsic LLRs

λe,det vector of extrinsic LLRs computed by the detector λe,dec vector of extrinsic LLRs computed by the decoder

λpi,b a posteriori LLR for the b-th bit on antenna i λai,b a priori LLR for the b-th bit on antenna i λei,b extrinsic LLR for the b-th bit on antenna i

R MIMO channel matrix after QRD

Ri,j element (i, j)of the MIMO channel matrix after QRD

˜

˜yi complex scalar received symbol after QRD on antenna i si complex modulated scalar symbol on transmit antenna i

s(i) partial transmit symbol vector from antenna i to antenna MT

xi,b binary label for the b-th bit of the complex modulated scalar symbol si

on antenna i

xi bit label for the complex modulated scalar symbol si on antenna i x bit label for the complex modulated transmit symbol vector s

Xi,b(±1) set of symbol vectors with the b-th bit on antenna i set to ±1

sML ML symbol vector output by the detector xMLi,b binary label for the b-th bit on antenna i of sML

xML bit label for the ML symbol vector sML

sMLi,b counter-hypothesis symbol vector for the b-th bit on antenna i of sML

sMAP MAP symbol vector output by the detector xMAPi,b binary label for the b-th bit on antenna i of sMAP

xMAP bit label for the MAP symbol vector sMAP

sMAPi,b counter-hypothesis symbol vector for the b-th bit on antenna i of sMAP

M(Ci) partial channel-based metric for the scalar symbol si on antenna i

MC(s) channel-based metric for symbol vector s

MC



s(i) channel-based metric for partial symbol vector s(i)

MML

C channel-based metric of the ML solution sML

M(Ai) partial a priori-based metric for the scalar symbol si on antenna i

MA(s) a priori-based metric for symbol vector s

MAs(i) a priori-based metric for partial symbol vector s(i)

M(Pi) partial general metric for the scalar symbol si on antenna i

MP(s) general metric for symbol vector s

MP



s(i) general metric for partial symbol vector s(i)

MMAPP general metric of the MAP solution sMAP

s(ik) symbol candidate selected by enumeration on antenna i at step k, i.e., after (k−1) nodes have been examined on antenna i

s(A,ik) symbol candidate selected by MA-based enumeration on antenna i at step k, i.e., after (k−1) nodes have been examined on antenna i

s(C,ik) symbol candidate selected by MC-based enumeration on antenna i at step k, i.e., after (k−1) nodes have been examined on antenna i

r2 sphere radius constraint for the tree search

Λe maximum absolute value allowed for clipped extrinsic LLRs Γ clipping value, normalised from Λe according to definition (4.4)

Γcaesar clipping value, normalised from Λe according to definition (3.4) Nen number of examined nodes in a single detector run

Nen,i number of examined nodes (same as Nen) in the i-th IDD iteration Nen,c number of examined nodes cumulated over multiple detector runs, i.e.,

over multiple IDD iterations

Nen,max maximum number of examined nodes allowed in a single detector run Nen,max(i) maximum number of examined nodes allowed on antenna i

ˆ

Λe correction vector for clipped extrinsic LLRs

ˆ Λe

i k-th entry of the correction vector ˆΛ

e ˆλe

i,b corrected extrinsic LLR for the b-th bit on antenna i

Λp LLR threshold for symbol-wise on-demand detection

Λe magnitude of the new LLR for symbol-wise on-demand detection i index of a given symbol vector within all the vectors in a codeword

N number of symbol vectors in a codeword C codebook of an error-correcting code Ni number of information bits in a codeword

Nc number of coded bits in a codeword (i.e., codeword length)

R code rate

Gb LDPC code generator matrix

Hb LDPC code parity-check matrix

Z sub-block size of IEEE 802.11n LDPC codes

Mp number of rows of an IEEE 802.11n LDPC matrix prototype Np number of columns of an IEEE 802.11n LDPC matrix prototype

Hp IEEE 802.11n LDPC matrix prototype

Pc cyclic-shift matrix

ck k-th coded bit in the codeword

λpk a posteriori LLR for the k-th bit in the codeword qv,c Message from variable node v to check node c

rc,v Message from check node c to variable node v

Nvn(c) set of neighbouring variable nodes to check node c Ncn(v) set of neighbouring check nodes to variable node v

β offset for correcting the messages rc,v in the OMS decoding algorithm

Idec number of LDPC iterations

B bandwidth in Hz

Bs symbol rate in sym/s

Bs,min minimum symbol rate requirement in sym/s Bs,max maximum symbol rate constraint in sym/s

Θ

i ideal information (or coded) throughput achievable in the absence of errors, in bit/s

Θi,min minimum ideal information (or coded) throughput achieved at symbol rate Bs,min, in bit/s

Θi,max maximum ideal information (or coded) throughput achieved at symbol rate Bs,max, in bit/s

ηs spectral efficiency in bit/s/Hz

Nf total number of decoded codewords (or frames)

Notation – Integrated Circuits

nemu number of parallel instances of a certain component implemented in a given FPGA-based emulator

MT,max maximum number of transmit antennae supported by a given architec- ture

Qmax maximum number of coded bits per complex modulated scalar symbol supported by a given architecture

CC cycle count for a given task fclk operating clock frequency in Hz fmax maximum clock frequency in Hz Vdd operating supply voltage in V Vdd,n nominal supply voltage in V Vth threshold voltage in V Is average static current in A Ps average static power in W Pd average dynamic power in W

S feature size ratio between two silicon technologies

U nominal supply voltage ratio between two silicon technologies Acore area occupied by one SD core in the Caesar chip in GE

Achip core area occupied by the Caesar chip in GE

Ps,core average static power consumed by one SD core in the Caesar chip in W Ps,chip average static power consumed by the Caesar chip in W

Θcaesar information (or coded) throughput of the Caesar architecture in bit/s

Gcaesar goodput of the Caesar architecture in bit/s Acaesar area occupied by the Caesar architecture in GE ηa,caesar area efficiency of the Caesar architecture in bit/s/GE

Pcaesar average total power consumed by the Caesar architecture in W ηe,caesar energy efficiency of the Caesar architecture in bit/J

CCdetc cycle count for detecting a complete codeword (cumulated over IDD iterations)

CCdetc, ideal ideal cycle count for detecting a complete codeword on a multicore detector (cumulated over IDD iterations)

CCdetmax maximum cycle count allowed for detecting a complete codeword (sin- gle IDD iteration)

CCdecit cycle count for one internal decoding iteration

CCdec cycle count for decoding a complete codeword (single IDD iteration) Aidd area of the IDD receiver in GE

fmax,det maximum clock frequency of the MIMO detector in Hz fmax,dec maximum clock frequency of the LDPC decoder in Hz

Θpe information (or coded) throughput of a processing element in bit/s Θ

det information (or coded) throughput of the MIMO detector in bit/s Θdec information (or coded) throughput of the LDPC decoder in bit/s Θidd information (or coded) throughput supported by the IDD receiver hard-

ware implementation in bit/s

Θidd,c information (or coded) throughput constrained by hardware imple- mentation and bandwidth requirements, in bit/s

Giterx goodput of the IteRX architecture in bit/s

Gc goodput constrained by hardware implementation and bandwidth re- quirements in bit/s

ηs,c spectral efficiency constrained by hardware implementation and band- width requirements in bit/s/Hz

ηa,idd area efficiency of the receiver in bit/s/GE Ps,idd static power consumed by the receiver in W Pd,det dynamic power consumed by the detector in W Pd,dec dynamic power consumed by the decoder in W ρpe utilisation ratio of a processing element

ρdet utilisation ratio of the MIMO detector ρdec utilisation ratio of the LDPC decoder

Es,idd static energy consumed by the IDD receiver while processing Nf code- words, in J

Ed,det dynamic energy consumed by the detector for processing Nf code- words, in J

Ed,dec dynamic energy consumed by the decoder for processing Nf code- words, in J

Eidd total energy consumed by the IDD receiver for processing Nf code- words, in J

Tdet total detector execution time for processing Nf codewords, in s Tdec total decoder execution time for processing Nf codewords, in s