{"paper":{"title":"Efficient Implementation of an Adaptive Transformer Accelerator for Massive MIMO Outdoor Localization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"An FPGA accelerator for adaptive Transformer-based 5G massive MIMO localization skips low-energy beams row-wise to deliver roughly 2x speedup with under 10% accuracy loss.","cross_cats":[],"primary_cat":"cs.AR","authors_text":"Ilayda Yaman, Liang Liu, Ove Edfors, Sijia Cheng","submitted_at":"2026-05-13T13:27:33Z","abstract_excerpt":"We present the implementation of an adaptive Transformer-based localization system for 5G massive MIMO targeting sub-10ms real-time positioning. The design exploits propagation characteristics, where beam-delay channel representations exhibit sparsity, enabling a row-wise skipping mechanism that removes low-energy beam components with minimal control overhead. The contribution is focused on hardware realization of the model using a mixed dataflow architecture, combining input- and output-stationary execution, mapped onto a heterogeneous vector processing engine with parallel processing element"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"The design achieves up to 65% row sparsity, yielding peak computational speedups of approximately 2x while limiting the average localization accuracy degradation to below 10%, relative to the floating-point baseline model. The accelerator attains below 1.15m localization accuracy across scenarios, with inference latency of 0.51-2.11ms and throughput of up to 1961 positions/s.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That real-world beam-delay channel representations exhibit sufficient and stable sparsity to allow row-wise skipping with only minor accuracy impact, and that the single-layer perceptron router provides reliable, low-latency model selection without introducing instability across environments.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"An FPGA accelerator for a sparsity-exploiting adaptive Transformer achieves up to 2x speedup and sub-2ms latency for massive MIMO localization with under 10% accuracy loss on real measurements.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"An FPGA accelerator for adaptive Transformer-based 5G massive MIMO localization skips low-energy beams row-wise to deliver roughly 2x speedup with under 10% accuracy loss.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"148b08001718bc4c91100294a9bb248363873a4f0719de05230692035e342b3c"},"source":{"id":"2605.13507","kind":"arxiv","version":1},"verdict":{"id":"9da24709-4061-4981-9354-0ce6efed0cb7","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T18:24:11.076213Z","strongest_claim":"The design achieves up to 65% row sparsity, yielding peak computational speedups of approximately 2x while limiting the average localization accuracy degradation to below 10%, relative to the floating-point baseline model. The accelerator attains below 1.15m localization accuracy across scenarios, with inference latency of 0.51-2.11ms and throughput of up to 1961 positions/s.","one_line_summary":"An FPGA accelerator for a sparsity-exploiting adaptive Transformer achieves up to 2x speedup and sub-2ms latency for massive MIMO localization with under 10% accuracy loss on real measurements.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That real-world beam-delay channel representations exhibit sufficient and stable sparsity to allow row-wise skipping with only minor accuracy impact, and that the single-layer perceptron router provides reliable, low-latency model selection without introducing instability across environments.","pith_extraction_headline":"An FPGA accelerator for adaptive Transformer-based 5G massive MIMO localization skips low-energy beams row-wise to deliver roughly 2x speedup with under 10% accuracy loss."},"references":{"count":28,"sample":[{"doi":"","year":2022,"title":"Service requirements for the 5G system,","work_id":"34cbedbb-8e5f-477b-81f4-f7f8487406d4","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2021,"title":"Study on NR positioning enhancements,","work_id":"8140612d-baf3-4a07-a53c-5305551d190d","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Attention-Aided Outdoor Localization in Commercial 5G NR Sys- tems,","work_id":"795df606-4b09-45b1-8563-9681d6c79d6d","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Adaptive Attention-Based Model for 5G Radio-Based Outdoor Localization,","work_id":"11b53f3b-eb10-44ef-90f7-eef5d7395449","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2019,"title":"A survey on 5G massive MIMO localization,","work_id":"86334556-b30b-4a17-b658-0a9856e955d2","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":28,"snapshot_sha256":"2a3445548217468d92798097bd098df92993932eacb1f32fc856d537ae838b5a","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}