| Recently, radiolocation has gained considerable attention from both academia and industry and becomes an indispensable application in wireless cellular/sensor networks. Positioning is required for purposes of routing, surveillance, public safety and military communications. Although GPS can provide location information, its availability is limited in urban settings, indoor, and in jammings. Here, we investigate multiuser/multipath channel estimation algorithms to obtain time of arrival (TOA) and if possible angle of arrival (AOA) for radiolocation. The geometric interpretation of the EM algorithm for multiuser channel estimation leads us to generalized successive interference cancellation (GSIC). GSIC appears to be a practical way to eliminate multiple access interference (MAI) in delay estimation and offers quadratic complexity instead of the exponential complexity of ML solution.; Combining matching pursuits (MP) with GSIC provides an efficient method to estimate a multiuser sparse channel. By constraining the number of nonzero channel coefficients, MP effectively avoids over-parameterization from which a least squares solution suffers. The estimates of TOA and AOA are then used in a distributed positioning algorithm, and positioning performance in dense urban environments is shown to be comparable to civilian GPS. Blind sparse channel estimation algorithms are also developed based on alternating maximization (AM) and per-survivor processing (PSP).; High resolution TOA estimation and channel impulse response order estimation are considered to solve the weak direct path detection problem. A modification of the minimum description length (MDL) penalty term is suggested by studying the optimal penalty criterion. A high-resolution MP algorithm developed by increasing the number of reference waveforms offers superior performance to MUSIC-type TOA estimation algorithms. Finally, a high resolution MP algorithm is combined with modified MDL to demonstrate the usefulness of the order estimation algorithm. |
