This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. 2 ||| 1556-6072/19©2019IEEE IEEE VEHICULAR TECHNOLOGY MAGAZINE | MARCH 2019 T he majority of existing wireless communication systems relies on training-based signaling, i.e., schemes that rely on accurate channel estimates for detecting the transmitted symbols. Training requires its own hardware and algorithms and consumes a significant portion of the channel coherence interval, which is typically short in vehicular communications. The drawbacks of training can be alleviated by using non- coherent signaling schemes, which dispense with the training phase altogether. We will focus on a particular class of such schemes, i.e., Grassmannian signaling. This scheme is optimal for high signal-to-noise ratio (SNR) communication over noncoherent block-fading channels, which are likely to arise in many future networks. Training-based signaling requires the receiver to ob- tain reliable channel estimates prior to detecting the transmitted symbols [1]. However, when the number of wireless nodes becomes large, as in the prospective par- adigms of machine-type communications (MTCs) and the Internet of Things (IoT), acquiring channel estimates constitutes a formidable task, especially for multiple- input, multiple-output (MIMO) systems. For these para- digms, a viable alternative is offered by noncoherent signaling, which does not require the receiver to have access to channel estimates. In comparison with their training-based counterparts, noncoherent schemes feature several advantages. First, they dispense with the hardware and the algorithms re- quired for learning the channel. To appreciate this, we note that, for the urban microcell (UMi) long-term evolu- tion (LTE) scenario of a user terminal operating at 1.8 GHz and traveling at 25 km/h, the channel coherence time is about eight symbol durations of 71.4 s n each. Using a four-antenna 3rd Generation Partnership Project training-based system to communicate in this scenario consumes 14.3% of the resources for training [2]; a cost that can be avoided by noncoherent signaling. Second, noncoherent schemes offer the potential of achieving higher spectral efficiencies. We will elaborate on this aspect later in this article. Third, noncoherent signaling alleviates one of the key problems that arise in massive MIMO cellular systems. In these systems, each base sta- tion (BS) is equipped with a large number of antennas, whereas the users are each equipped with a small num- ber of antennas. When training-based schemes are used in the uplink of these systems, multiple neighboring cells will reuse the same training pilots. The received signals corresponding to these pilots interfere, resulting in pilot contamination. Hence, using noncoherent signaling in Digital Object Identifier 10.1109/MVT.2018.2866306 Date of publication: 4 January 2019 NONCOHERENT MIMO SIGNALING FOR BLOCK- FADING CHANNELS Approaches and Challenges Ramy H. Gohary and Halim Yanikomeroglu