A context-aware cognitive SIMO transceiver for enhanced throughput on the downlink of LTE HetNet

In this paper, we design a new single-input multiple-output context-aware cognitive transceiver (CTR) that is able to switch to the best performing modem in terms of link-level throughput. On the top of conventional adaptive modulation and coding, we allow the proposed CTR to make best selection between three different pilot-utilization modes: conventional data-aided (DA) or pilot-assisted, non-DA (NDA) or blind, and NDA with pilots, which is a newly proposed hybrid version between the DA and NDA modes. We also enable the CTR to make best selection between two different channel identification schemes: conventional least-squares (LS) and newly developed maximum-likelihood estimators. Depending on whether pilot symbols can be exploited or not at the receiver, we further enable the CTR to make the best selection among two data detection modes: coherent or differential. Owing to exhaustive link-level simulations on the downlink of a long-term evolution system, we draw out the optimal decision rules in terms of the best combination triplet of pilotuse, channel-identification, and data-detection modes that yield the best link-level throughput as function of channel type, mobile speed, signal-to-noise ratio, and channel quality indicator. The proposed CTR offers a link-level throughput gains improvement as high as 700% compared with DA LS for VehB channel type at a mobile speed of 100 km/h in the low signal-to-noise ratio region. For VehA channels, its throughput gain improvements can reach 114%. For PedA and PedB channels, the proposed CTR provides throughput enhancements of about 66% and 330%, respectively. Moreover, realistic simulations at the system-level of the long-term evolution-HetNet network suggest that the new context-aware CTR outperforms the conventional transceiver (i.e., pilot-assisted LS-type channel estimation with coherent detection) by as much as 50% and 60% gains in average and cell-edge (i.e., five percentile) throughputs, respectively, in the high-clustering case with type-B channels. In the low clustering scenario, the average and cell-edge throughput gain improvements offered by the proposed CTR exceed 80% and 90% for type-B channels. Copyright © 2016 John Wiley & Sons, Ltd.