Abstract:

The complex nature of the mobile wireless channel has fueled research in recent years to discover ways to either combat the disadvantages or exploit the benefits of the fading channel. As a result, new paradigms such as multiple input multiple output (MIMO) processing, and orthogonal frequency division multiplexing (OFDM) have flourished based on underlying theoretical concepts such as adaptation and diversity in space, time, frequency or multiuser dimensions.

In this dissertation the effects that channel knowledge has on the theoretical and practical limits of wireless communications are ivestigated.  First, the impact of channel state information (CSI) on the outage capacity of a single input single output (SISO) Rayleigh flat fading channel, completely unknown at the transmitter and partially known at the receiver, is studied. Analytical expressions for the upper and lower bounds of the outage capacity are obtained as a function of the channel estimation error at the receiver. Those results reveal how the noise level and the channel uncertainty combine in reducing the outage capacity. In addition, by deriving an expression for the estimation error variance of pilot symbol assisted modulation (PSAM), we are able to establish direct relations among estimation parameters, channel characteristics (in terms of SNR and normalized Doppler spread) and its impact on the Outage Capacity.

Secondly, a new approach for the design of practical adaptive modulation and coding (AMC) for single user MIMO-OFDM communications is introduced. A possible conflict on the joint extraction of diversity and adaptation benefits of a MIMO-OFDM channel when full CSI is available at the transmitter is justified. The proposed scheme avoids this conflict by adapting as accurately as possible given the side information while using diversity to combat remaining uncertainty. This is accomplished by channel parallelization, sorting and grouping into modulation modes. A new power allocation strategy, named water-filling/channel-inversion (WF/CI), is also introduced and shown to perform better than conventional water-filling in a practical setting of finite modulation and coding modes.

For the multiuser scenario, a novel adaptive transmission scheme for the downlink is presented; where the base station (BS) is equipped with M transmit antennas. It is shown that very few bits of feedback provide the BS with the sufficient CSI to capture most of the multiuser diversity. The proposed scheme is used to prove that the optimal scaling law of the sum-rate of the MIMO broadcast channel with full CSI (Mloglog nN where n is the number of users and N the number of receive antennas), can be also achieved with only one bit of feedback per user. The impact of increasing the feedback load further from 1 bit per user is also assessed. An asymptotic analysis of the multiplexing gain leads to practical design guidelines for the proposed scheme for relatively low number of users. Finally, for the case of N>M, a new technique that employs linear receivers and only one bit of feedback per user is derived, and its performance analyzed.

Committee members:

Prof. Yeheskel Bar-Ness, Dept. of Electrical and Computer Engineering, NJIT. Advisor and committee chair

Prof. Alexander Haimovich, Dept. of Electrical and Computer Engineering, NJIT.

Prof. Ali Abdi, Dept. of Electrical and Computer Engineering, NJIT.

Prof. Hongya Ge, Dept. of Electrical and Computer Engineering, NJIT.

Prof. Roy Yates, Dept. of Electrical and Computer Engineering, Rutgers University.