In　order　to　achieve　more　efficient　and　accurate　DDoS　detection　while　ensuring　data　privacy,　this　paper　proposes　a　DDoS　detection　method　based　on　FLAD.　Firstly,　this　paper　uses　the　FLAD　algorithm　to　train　a　global　DDoS　detection　model　without　leaving　local　traffic　data,　protecting　the　privacy　and　security　of　traffic　data　between　different　hosts,　and　improving　aggregation　efficiency　by　dynamically　adjusting　the　aggregation　weights　to　adapt　to　different　sub-dataset　increments.　Secondly,　a　DDoS　traffic　detection　method　based　on　the　integration　of　LSTM　and　CNN　is　proposed,　which　extracts　and　analyzes　the　temporal　correlation　of　traffic　data　by　calculating　the　statistical　characteristics　of　traffic　data　within　a　time　period,　to　achieve　real-time　detection　of　traffic　feature　data.　Again,　combined　with　the　concept　of　SDN,　real-time　defense　against　DDoS　based　on　ODL-API　is　implemented,　and　precise　matching　of　DDoS　detection　results　with　network　entity　information　is　achieved,　realizing　the　technology　of　real-time　and　precise　issuance　of　multiple　flow　rules,　effectively　blocking　DDoS　malicious　attack　traffic,　protecting　important　entities　in　the　topology,　and　maintaining　stable　traffic　in　the　topology.　This　paper　focuses　on　solving　the　detection　problem　of　DDoS　traffic　data　increments　and　uneven　data　distribution　through　the　FLAD　algorithm.　Experimental　results　show　that　the　proposed　method　improves　the　accuracy　of　DDoS　attack　detection　by　more　than　4%　and　the　F1　Score　by　more　than　7%　compared　to　the　FedAvg　aggregation　algorithm.　©　2024　IEEE.