Joint channel and frequency offset estimation in CDMA systems

The problem of frequency offset and channel estimation is essential for multiuser detectors in a DS-CDMA system. An efficient algorithm for joint estimation of these parameters without using training sequences is proposed. The approach exploits the ESPRIT algorithm by collecting two samples within each chip duration and constructing an approximate shift invariance structure. The high-resolution and noise-robust frequency offset estimation can therefore be achieved. Although the proposed algorithm is suboptimal relative to the maximum likelihood estimators, it provides the closed-form solutions and much less complexity in computation