Multiplex communications – Generalized orthogonal or special mathematical techniques – Fourier transform
Reexamination Certificate
2000-01-20
2004-01-20
Pham, Chi (Department: 2667)
Multiplex communications
Generalized orthogonal or special mathematical techniques
Fourier transform
C370S335000, C370S342000, C370S441000, C375S130000, C375S146000, C375S147000, C375S229000
Reexamination Certificate
active
06680902
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to a spreading code selection process for wireless communications systems such as CDMA systems. The present invention relates, in particular, but is not necessarily limited, to third generation TDD, and FDD cellular systems, data systems therefore and very high speed downstream internet access CDMA systems.
BACKGROUND OF THE INVENTION
Cellular wireless has enjoyed an extremely rapid growth from the 1980s. There is now an almost widespread coverage of cellular radio services in industrialised countries. In recent years there has been a similar, albeit more rapid growth in the demand for data services from systems such as the internet, intranet and other data transmission systems. People have become familiar with the advantages of both mobile wireless voice communications and web speed data communications. A demand is growing for mobile wireless data services. The demand for such services is illustrated by the extent of e-mail retrieval, web browsing etc. that already takes place in such locations as airport lounges, hotel lobbies, company conference rooms, etc.
Data traffic is assymmetric, in contrast with that voice traffic. A far greater data rate is necessary for a down link from a wireless access point to a subscriber terminal than that required for of the reverse link (or uplink).
One proposed wireless data system Is the High speed data (HSD) system which provides a high speed, high capacity wireless technologies compatible with CDMA networks for data services. It is intended that the HSD system will require minimal network and spectrum resources. The proposed HSD system provides a shared resource architecture rather than a circuit switched system whereby network, spectrum and air-link resources are minimised.
HSD follows on from existing CDMA technologies such as the IS-95 systems. The RF characteristics of IS-95 are examined whereby operators may make evolutionary changes to their existing IS-95 system. Existing network may be retained, i.e. components cost levels may be reduced by the continued use of existing technology components and devices. Some HSD proposals decouple data services from voice services: rather than providing an equal grade of services to all users. HSD proposes to allocate each user a maximum data rate possible, dependant upon application requirements and wireless channel conditions.
In the TDD mode of UTRA, as standardised in the 3
rd
generation partnership (3GPP) the downlink channels can use one code with a spreading factors of 1, 2, 4, 8, 16. The parallel spreading factors are orthogonal and
FIG. 1
represents an orthogonal variable spreading factor (OVSF). This allows the possibility of mixing the spreading factors if a user has several different channels open simultaneously. Prior to this the SF would be fixed at 16. Control signals will use their own fixed spreading factors, irrespective of the SF chosen for the data.
A further proposal is that certain mixed spreading factors can be employed whereby a “multicode” extension can be employed which allows one user to have several code words in order to offer a fine grained set of bit rates.
In the situation where all branches of the orthogonal spreading factor (OVSF) tree are in use, given that the simultaneous use of high and low SF's on the same branch is forbidden due to non-orthogonality, the optimum downlink receiver is essentially the zero-forcing (ZF) filter or its MMSE variant. If some branches of the OVSF tree are not in use then the ZF solution is suboptimum and higher CNR can be developed by using a ZF solution optimised for the few spreading codes with the highest common SF. Block based solutions to the uplink problems in TDD are essentially the same as the down-link case though each user has a different channel impulse response which Increases the work load.
It has been shown that the uplink FDD component of CDMA can be equalised and multiple access interference cancelled using what is essentially a TDD algorithm. In order to do this it was assumed that the short scrambling codes option would be used and the block equations are written out with a block size equal to the scrambling code length of 256 bits, which, whilst giving a high work load, is practicable. With some numerical Improvements to some known methods it becomes possible to consider applying a least squares linear solution for the long scrambling code FDD uplink option also. The resources available for solving least squares linear equations are matrix inversion, Cholesky triangularisation and back substitution, block FFT's, circulants and pseudoinverse methods. However there are unlimited ways for combining theses methods and for factoring and partitioning of the sets of equations in order to get down to the irreducible matrix which must be inverted.
Code words are selected from the orthogonal variable spreading factor (OVSF) tree which allows code words of different length to be mixed yet remain orthogonal. If a code passes through a node A in this tree ten no other shorter code word can be used which passes through node A. Thus when the code word C
4,0
={1,1,1,1} is in use, another user would be allowed to use word C
4,1
={1,1,−1,−1} which Is orthogonal but another user would not be allowed to use word C
2,0
{1,1} which, if repeated twice: {1,1},{1,1}, would not be orthogonal to C
4,0
.
Equalisation in terminals is common for the TDD mode of CDMA as it enhances BER performance and can increase the spectral efficiency. Low cost DSP equalisers are available for TDD systems. There are several types of equaliser which have low work load and are suitable for use in battery powered terminals but which can only provide a restricted DSP capability due to battery power drain considerations. These are:
(i) Cholesky factorisation of the channel impulse response autocovariance matrix. This factorisation is efficient when the autocovariance matrix is strongly banded (associated with low dispersion channels). After matrix factorisation the least squares filter equations are solved by back substitution as usual in the Cholesky method.
(ii) Decision feedback equaliser (DFE). This uses a combination of forward FIR filtering, a threshold decision device, and a feedback filter and is commonly used in FDMA applications such as the US 2
nd
generation cellular phone receivers and telephone modem equalisers.
(iii) Zero-forcing filter. Here an FIR fitter is synthesised which equalises the channel dispersion for a finite time span about the origin. For an n-tap FIR filter solution to a set of linear equations will always form a weight set such that the convolution of the sample channel impulse response h
k
and the filter w
k
will have a combined impulse response which is zero at n−1 arbitrary points and has unit response at k=0.
(iv) The Wiener least squares filter. This is a modified inverse filter which controlled the white noise response of the filter, ie. the undesired enhancement of thermal noise from the antenna. If the discrete frequency response of the channel is H
k
and the thermal noise variance is &sgr;
2
then the Wiener filter frequency response is
W
k
=
H
k
*
&LeftBracketingBar;
H
k
&RightBracketingBar;
2
+
σ
2
A Fast Fourier transform can be used to form this solution with low average work load and this technique is disclosed in a separate patent application, filed concurrently with this application.
In general least squares solutions such as (i) and (iv) are better than algebraic solutions such as zero forcing solutions and give better BER performance. The optimal least square filter is a function of the number of code words in use and their spreading factors. For example a LS filter selected to be optimum for use with a subset of the codes, as is common in 2
nd
generation CDMA systems, is different from one which is optimum for the case that all codes are in use at the same time. It Is also much more complicated to compute the filter coefficients for use with a cod
Barnes & Thornburg
Nortel Networks Limited
Pham Chi
Qureshi Afsar M.
LandOfFree
Spreading code selection process for equalization in CDMA... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Spreading code selection process for equalization in CDMA..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Spreading code selection process for equalization in CDMA... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3229559