This　paper　concerns　the　detection　of　Distributed　Denial　of　Service　(DDoS)　attacks　in　network　traffic　generated　by　Internet　of　Things　(IoT)　devices　in　smart　home　environments.　The　detection　of　DDoS　attacks　is　crucial　for　IoT　network　security,　as　such　attacks　can　disrupt　the　availability　of　essential　services.　In　particular,　due　to　the　growing　popularity　of　smart　homes　and　the　emergence　of　malicious　software　that　compromises　devices,　home　IoT　devices　have　become　susceptible　to　botnet　infections　capable　of　launching　DDoS　attacks.　With　the　development　of　artificial　intelligence　technology,　many　advanced　methods　have　been　proposed　that　show　promising　performance　in　detecting　DDoS　attacks.　However,　there　is　still　a　need　for　improvement　in　their　generalizability　and　detection　efficiency.　In　this　paper,　we　propose　Hifoots,　a　highly　efficient　IoT　DDoS　attack　detection　scheme,　aiming　to　achieve　high　detection　robustness　and　detection　efficiency.　Hifoots　builts　upon　our　key　observation　that　DDoS　attacks　can　be　detected　by　examining　the　group　behavior　of　all　flows　over　a　given　time　interval.　We　evaluated　Hifoots　on　five　complex　DDoS　attack　scenarios.　The　experimental　results　demonstrate　that　Hifoots　outperforms　the　detection　performance　of　existing　state-of-the-art　methods　and　offers　an　improvement　in　detection　efficiency　that　is　up　to　12　times　better,　along　with　stronger　generalizability　compared　to　the　state-of-the-art　methods.　IEEE