ACCsiNet: Asymmetric Convolution-Based Autoencoder Framework for Massive MIMO CSI Feedback

Biao Cao,Yang Yang,Peng Ran,Dazhong He,Gang He
DOI: https://doi.org/10.1109/lcomm.2021.3116864
IF: 3.5529
2021-12-01
IEEE Communications Letters
Abstract:Channel state information (CSI) is a critical part for massive multiple-input multiple-output (MIMO) system. However, it is a big challenge to send a large amount of CSI from the receiver to the transmitter with limited channel resources. In this letter, we propose asymmetric convolution-based autoencoder framework (ACCsiNet) to handle the CSI compression and decompression problem. Specifically, asymmetric convolution block (AC-Block) is used to enhance the feature extraction ability of convolution. Further, a lightweight method is applied, which can greatly reduce the storage space at the receiver. Considering the practical deployment, multi-model fusion schemes including multi-rate and multi-scenario fusion are also discussed to strengthen the generalization ability of the network. Experimental results show that the proposed ACCsiNet can improve the NMSE and cosine similarity <span class="mjpage"><svg xmlns:xlink="http://www.w3.org/1999/xlink" width="3.515ex" height="2.176ex" style="vertical-align: -0.338ex;" viewBox="0 -791.3 1513.5 936.9" role="img" focusable="false" xmlns="http://www.w3.org/2000/svg"><g stroke="currentColor" fill="currentColor" stroke-width="0" transform="matrix(1 0 0 -1 0 0)"> <use xlink:href="#MJMATHI-72" x="0" y="0"></use> <use xlink:href="#MJMATHI-68" x="451" y="0"></use> <use xlink:href="#MJMATHI-6F" x="1028" y="0"></use></g></svg></span> performance, especially for outdoor scenario. The results also verify that both the lightweight and multi-model fusion schemes can reach a near-optimal performance of the proposed ACCsiNet, but further significantly reduce the parameter amount by more than 83% and 90%, respectively.<svg xmlns="http://www.w3.org/2000/svg" style="display: none;"><defs id="MathJax_SVG_glyphs"><path stroke-width="1" id="MJMATHI-72" d="M21 287Q22 290 23 295T28 317T38 348T53 381T73 411T99 433T132 442Q161 442 183 430T214 408T225 388Q227 382 228 382T236 389Q284 441 347 441H350Q398 441 422 400Q430 381 430 363Q430 333 417 315T391 292T366 288Q346 288 334 299T322 328Q322 376 378 392Q356 405 342 405Q286 405 239 331Q229 315 224 298T190 165Q156 25 151 16Q138 -11 108 -11Q95 -11 87 -5T76 7T74 17Q74 30 114 189T154 366Q154 405 128 405Q107 405 92 377T68 316T57 280Q55 278 41 278H27Q21 284 21 287Z"></path><path stroke-width="1" id="MJMATHI-68" d="M137 683Q138 683 209 688T282 694Q294 694 294 685Q294 674 258 534Q220 386 220 383Q220 381 227 388Q288 442 357 442Q411 442 444 415T478 336Q478 285 440 178T402 50Q403 36 407 31T422 26Q450 26 474 56T513 138Q516 149 519 151T535 153Q555 153 555 145Q555 144 551 130Q535 71 500 33Q466 -10 419 -10H414Q367 -10 346 17T325 74Q325 90 361 192T398 345Q398 404 354 404H349Q266 404 205 306L198 293L164 158Q132 28 127 16Q114 -11 83 -11Q69 -11 59 -2T48 16Q48 30 121 320L195 616Q195 629 188 632T149 637H128Q122 643 122 645T124 664Q129 683 137 683Z"></path><path stroke-width="1" id="MJMATHI-6F" d="M201 -11Q126 -11 80 38T34 156Q34 221 64 279T146 380Q222 441 301 441Q333 441 341 440Q354 437 367 433T402 417T438 387T464 338T476 268Q476 161 390 75T201 -11ZM121 120Q121 70 147 48T206 26Q250 26 289 58T351 142Q360 163 374 216T388 308Q388 352 370 375Q346 405 306 405Q243 405 195 347Q158 303 140 230T121 120Z"></path></defs></svg>
telecommunications
What problem does this paper attempt to address?