We study the blind symbol estimation problem in digital communications
and propose a novel algorithm by exploiting a special data structure
of an oversampled system output, Unlike most equalization schemes that
involve two stages-channel identification and channel equalization/sy
mbol estimation-the proposed approach accomplishes direct symbol estim
ation without determining the channel characteristics. Based on a dete
rministic model, the new method can provide a closed-form solution to
the symbol estimation using a small set of data samples, Which makes i
t particularly suitable for wireless applications with fast changing e
nvironments, Moreover, if the symbols belong to a finite alphabet, e.g
., BPSK or QPSK, our approach can be extended to handle the symbol est
imation for multiple sources, Computer simulations and field RF experi
ments were conducted to demonstrate the performance of the proposed me
thod, The results are compared to the Cramer-Rao lower bound of the sy
mbol estimates derived in this paper.