Method for calculating functions of the channel matrices in linear MIMO-OFDM data transmission
AssigneeEth Zurich, Eth Transfer
InventorsAndreas Burg, Helmut Bölcskei, Moritz Borgmann, Davide Cescato, Jan Hansen
Life of this patent
Estimating the term…
Abstract
The present application describes a method for transmitting data through a MIMO-OFDM system which requires the computation of a function f(H(sk)), where H is the channel matrix and sk=exp(j*2*pi*k/N). The method splits function f into a first, nonpolynomial part, which is in turn a function of one or more polynomial functions in sk.
Claims16
A method for transmitting data through a MIMO-OFDM system having a transmitter and a receiver that communicate through a channel, wherein the transmitter comprises MT transmitter units TU0 . . . TUMT−1 and the receiver comprises MR receiver units RU0 . . . RUMR−1, said method comprising the steps of
- a) feeding a sequence of original symbols to said transmitter, each original symbol comprising N>1 vectors ck with k=0 . . . N−1, wherein each vector ck has MT complex-valued coordinates (c0)k . . . (cMT−1)k to be fed to the transmitter units TU0 . . . TUMR−1,
- b) OFDM-modulating said original symbols in said transmitter units and transmitting them to said receiver units,
- c) receiving, by said receiver units, said transmitted symbols distorted by the channel and decoding them into received symbols, wherein each received symbol comprises N vectors rk with k=0 . . . N−1, wherein each vector rk has MR complex-valued coordinates (r0)k . . . (rMR−1)k, wherein, at least in approximation,
- rk=√{square root over (Es)}·H(sk)ck+wk,
- with
- √{square root over (Es)} being a constant,
- sk=exp(−j2πk/N), j being the imaginary unit,
- H
-
- (
- s
- k
- )
- =
- ∑
- l
- =
- 0
- L
- -
- 1
-
- H
- l
-
- s
- k
- l
- ,
- L an integer smaller than N,
- Hl being an MR×MT matrix having coefficients (Hij)l describing the value at discrete-time delay l of the discrete-time impulse response of the channel between transmitter unit TUj and receiver unit RUi, and
- wk being a noise vector,
- d) said method further comprising the step of carrying out at least one computation of a function f(H(sk)) for k=0 . . . N−1 on said matrices H(sk), which function f can be expressed as
- f(H(sk))=fNP(fP0(H(sk)), . . . fPJ−1(H(sk))) with J≧1 and scalar-, vector- or matrix-valued polynomial functions fP0(H(sk)), . . . fPJ−1(H(sk)) that can be expressed as
- f
- P
- i
-
- (
- H
-
- (
- s
- k
- )
- )
- =
- ∑
- l
- =
- 0
- N
- -
- 1
-
- A
- l
- i
-
- s
- k
- l
- with coefficients Ali
- wherein a number Qi of said coefficients Ali are arbitrary functions
- Ali=gli(H0, . . . HL−1) of the matrices H0, . . . , HL−1
- and a number N−Qi said coefficients Ali are zero, with 1≦Qi<N
- e) said method further comprising the step of carrying out at least one computation of a said function fPi(H(sk)) using the steps comprising
- step A) determining from known H(sk), for at least some of said polynomial functions fPi(H(sk)), a number of Vi≧Qi values of said function fPi(sk) at a set of sk with k∈Ωi and |Ωi|=Vi≧Qi, and
- step B) determining, for at least some of said polynomial functions fPi(H(sk)), at least one value fPi(H(sk)) for an sk with k∉Ωi by interpolation of the values of said function fPi(H(sk)) obtained in step A).
Dependencies are inferred from the claim text.