{"id":4828,"date":"2025-11-25T17:21:28","date_gmt":"2025-11-25T17:21:28","guid":{"rendered":"https:\/\/172-234-197-23.ip.linodeusercontent.com\/?page_id=4828"},"modified":"2025-11-25T17:21:28","modified_gmt":"2025-11-25T17:21:28","slug":"iq-length-normalization-policies-for-rf-modulation-classifiers","status":"publish","type":"page","link":"https:\/\/neurosphere-2.tail52f848.ts.net\/wordpress\/?page_id=4828","title":{"rendered":"IQ Length Normalization Policies for RF Modulation Classifiers"},"content":{"rendered":"\n<div data-wp-interactive=\"core\/file\" class=\"wp-block-file\"><object data-wp-bind--hidden=\"!state.hasPdfPreview\" hidden class=\"wp-block-file__embed\" data=\"https:\/\/neurosphere-2.tail52f848.ts.net\/wordpress\/wp-content\/uploads\/2025\/11\/IQ-Length-Normalization-Policies-for-RF-Modulation-Classifiers-Spectrcyde.pdf\" type=\"application\/pdf\" style=\"width:100%;height:600px\" aria-label=\"Embed of IQ Length Normalization Policies for RF Modulation Classifiers - Spectrcyde.\"><\/object><a id=\"wp-block-file--media-ceddee58-8b79-4889-bb2d-71c6d9b2efa3\" href=\"https:\/\/neurosphere-2.tail52f848.ts.net\/wordpress\/wp-content\/uploads\/2025\/11\/IQ-Length-Normalization-Policies-for-RF-Modulation-Classifiers-Spectrcyde.pdf\">IQ Length Normalization Policies for RF Modulation Classifiers &#8211; Spectrcyde<\/a><a href=\"https:\/\/neurosphere-2.tail52f848.ts.net\/wordpress\/wp-content\/uploads\/2025\/11\/IQ-Length-Normalization-Policies-for-RF-Modulation-Classifiers-Spectrcyde.pdf\" class=\"wp-block-file__button wp-element-button\" download aria-describedby=\"wp-block-file--media-ceddee58-8b79-4889-bb2d-71c6d9b2efa3\">Download<\/a><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">Temporal RF models typically require fixed-length IQ<br>sequences, yet real-world bursts arrive at variable durations and<br>sampling rates. In RF\u2013QUANTUM\u2013SCYTHE, the temporal input<br>builder _create_temporal_input normalizes each complex<br>IQ stream to a configured sequence length before feeding recurrent<br>and transformer-style encoders.<br>This paper compares three practical IQ length normalization<br>policies\u2014evenly spaced downsampling, windowed pooling, and<br>strided crops\u2014in a shared RF modulation classification stack.<br>We sweep sequence length from very short (tens of samples) to<br>long (hundreds to thousands) and quantify the trade-off between<br>aliasing distortion and classification accuracy. On synthetic RF scenarios, we find that simple evenly spaced downsampling achieves<br>near-baseline accuracy at modest lengths, while aggressive strided<br>cropping can shed computation but risks missing informative<br>structure. The windowed pooling policy provides a middle ground,<br>smoothing local variations at the cost of mild aliasing. On our<br>synthetic RF benchmark, evenly spaced downsampling retains<br>up to 89.2% accuracy at L=128, while more aggressive crops<br>and pools trade a few percentage points of accuracy for reduced<br>temporal resolution. We release a harness and figure-generation<br>scripts so new policies and lengths can be evaluated without<br>modifying the LATEX.<br>Index Terms\u2014Automatic modulation classification, RF machine<br>learning, IQ processing, sequence length, downsampling.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Repository:<\/strong> <a href=\"https:\/\/github.com\/bgilbert1984\/IQ-Length-Normalization-Policies-for-RF-Modulation-Classifiers\">bgilbert1984\/IQ-Length-Normalization-Policies-for-RF-Modulation-Classifiers: Temporal RF models typically require fixed-length IQ sequences, yet real-world bursts arrive at variable durations and sampling rates. In RF\u2013QUANTUM\u2013SCYTHE, the temporal input builder _create_temporal_input normalizes each complex IQ stream to a configured sequence length before feeding recurrent and transformer-style encoders.<\/a><\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><\/p>\n","protected":false},"excerpt":{"rendered":"<p>Temporal RF models typically require fixed-length IQsequences, yet real-world bursts arrive at variable durations andsampling rates. In RF\u2013QUANTUM\u2013SCYTHE, the temporal inputbuilder _create_temporal_input normalizes each complexIQ stream to a configured sequence length before feeding recurrentand transformer-style encoders.This paper compares three practical IQ length normalizationpolicies\u2014evenly spaced downsampling, windowed pooling, andstrided crops\u2014in a shared RF modulation classification stack.We&hellip;&nbsp;<\/p>\n","protected":false},"author":2,"featured_media":4830,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"neve_meta_sidebar":"","neve_meta_container":"","neve_meta_enable_content_width":"","neve_meta_content_width":0,"neve_meta_title_alignment":"","neve_meta_author_avatar":"","neve_post_elements_order":"","neve_meta_disable_header":"","neve_meta_disable_footer":"","neve_meta_disable_title":"","footnotes":""},"class_list":["post-4828","page","type-page","status-publish","has-post-thumbnail","hentry"],"_links":{"self":[{"href":"https:\/\/neurosphere-2.tail52f848.ts.net\/wordpress\/index.php?rest_route=\/wp\/v2\/pages\/4828","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/neurosphere-2.tail52f848.ts.net\/wordpress\/index.php?rest_route=\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/neurosphere-2.tail52f848.ts.net\/wordpress\/index.php?rest_route=\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/neurosphere-2.tail52f848.ts.net\/wordpress\/index.php?rest_route=\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/neurosphere-2.tail52f848.ts.net\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=4828"}],"version-history":[{"count":0,"href":"https:\/\/neurosphere-2.tail52f848.ts.net\/wordpress\/index.php?rest_route=\/wp\/v2\/pages\/4828\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/neurosphere-2.tail52f848.ts.net\/wordpress\/index.php?rest_route=\/wp\/v2\/media\/4830"}],"wp:attachment":[{"href":"https:\/\/neurosphere-2.tail52f848.ts.net\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=4828"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}