Signature sequence selection, system value bit loading and energy allocation method and apparatus for multicode single-input single-output and multiple-input multiple-output parallel channels

A method of transmitting data over a radio data transmission system having a plurality of K parallel single-input single-output or multiple-input multiple-output channels, the method comprising transmitting data at a rate bp bits per symbol over a first group of (K−m) channels, and at a rate bp+1 bits per symbol over a second group of m channels, by spreading the data using a number of signature sequences S;

• wherein the total number of signature sequences is greater than one, and is equal to the multiplication of the number of receiving antennas and the processing gain N used to spread the system signals;
• wherein the spreading signature sequences S are determined using the Gramian matrices Q=HHH of the channel impulse responses of the frequency selective multipath radio channels,
• where the channel impulse response matrix H is obtained by forming the matrix
• H
• =
• [
• H
• 1
• ,
• 1
• H
• 1
• ,
• 2
• H
• 2
• ,
• 1
• H
• 2
• ,
• 2
• ]
• using the specific channel impulse response matrix Hi,j which is defined as the multipath convolution matrix for a pair of transmitting antenna i and receiving antenna j, where i and j are integer numbers one or more,
• and where the signature sequences S are obtained by decomposing the Gramian matrix Q into its Eigen vectors V as Q=VDVH, where D is the matrix of Eigen values, and then by setting S=V;
• wherein the optimum number of transmission channels is identified by using the water filling method where the signature sequence matrix S is ordered such that the channel gains |hk|2, which are diagonal elements of D, appear in a descending order and the matched filter channel-SNIR gk for channel k is calculated using
• g
• k
• =
• 
• h
• k
• 
• 2
• 2
• 
• 
• σ
• 2
• for k=1, . . . , K where 2σ2 is the noise per channel for the system with
• σ
• 2
• =
• N
• 0
• 2
• for the two sided noise power spectral density of
• N
• 0
• 2
• ;
• and
• wherein the optimum number, K*, of the signature sequences to be used is identified by initially setting K* to be K*=K and by calculating the water filling energies
• E
• k
• =
• 1
• K
• *
• 
• [
• E
• T
• +
• 1
• Γ
• 
• ∑
• k
• =
• 1
• K
• *
• 
• 
• 1
• g
• k
• ]
• =
• 1
• Γ
• 
• 
• g
• k
• for k=1, . . . , K* and then by testing the energy EK*, for the last channel K*, to check if the energy is negative and for the negative energy case the optimum number K* is set to be (K*−1) and the energy calculation process is repeated until all energies are positive and for the resultant K* channels the signature sequences S=[{right arrow over (s)}1 . . . {right arrow over (s)}K*] are re-ordered such that the corresponding channel gains |hk|2 appear in an ascending order and the despreading sequence matrix is reorganized such that
• S
• =
• [
• s
• →
• 1
• 
• 
• …
• 
• 
• s
• →
• K
• *
• ]
• =
• [
• S
• 1
• S
• 2
• ]
• where the signature sequences given by the N×K* dimensional matrices S1=└{right arrow over (s)}1,1 . . . {right arrow over (s)}1,K*┘ are used to load the first K* spreading units attached to a first transmitting antenna and S2=└{right arrow over (S)}2,1 . . . {right arrow over (s)}2,K*┘ are used to load the second K* spreading units attached to a second transmitting antenna.