Channel Charting for Pilot Reuse in mMTC with Spatially Correlated MIMO Channels

Exploiting Spatial Correlation for Pilot Reuse in Single-Cell mMTC

“…CC is a framework proposed in [12] to estimate the relative position of devices in an unsupervised manner which maps the information obtained from the measured long-term CSI at the BS into a low-dimensional chart, in which the relative positions of UEs are preserved. Our results [11] showed that CSI can be utilized to allocate the pilot sequences, and improve the channel estimation accuracy and symbol error rate (SER), for single-cell correlated channels systems.…”

“…In particular, we propose two features encompassing the channel correlation, especially in the angular domain, for the pilot reuse: 1) the CMD metric and 2) the CC created based on the CMD metric. These features are fed to a modified version of our greedy pilot allocation algorithm proposed in [11] (Algorithm 1 detailed in Sect. V-C), which completes the pilot assignment, aiming at minimizing the pilot interference between the users sharing the same pilot.…”

“…Recently, we proposed a pilot reuse scheme in [11] that utilizes channel charting (CC) to exploit the spatial information existing in CSI, aiming to maximize the AoA distances between the UEs using the same pilot sequence. CC is a framework proposed in [12] to estimate the relative position of devices in an unsupervised manner which maps the information obtained from the measured long-term CSI at the BS into a low-dimensional chart, in which the relative positions of UEs are preserved.…”

“…To exploit the degree of orthogonality between the UEs' channels, granted by the spatial correlation, the pilot reuse is performed based on two features encompassing the channel correlation, especially in the angular domain: 1) the CMD metric and 2) the CC created based on the CMD metric. Utilizing the proposed features along with the pilot allocation algorithm devised in [11], the numerical results show that the angular information can be exploited also for multi-sector systems with spatially correlated channels, resulting in enhanced achievable rate and SER compared to baselines.…”

Efficient Channel Charting via Phase-Insensitive Distance Computation

“…The chart should reflect as much as possible the physical reality, in the sense that the charting function should preserve spatial neighborhoods. Predicting this way the relative locations of users from channel measurements has many potential applications, ranging from SNR prediction [3] and pilot reuse [4], to user grouping, proactive handover management or beam-finding (see [2] for more details on potential applications). What makes channel charting particularly interesting compared to classical positioning methods is its fully unsupervised nature.…”