Our　previous　sparse　regularization　method　for　multipath　interference　(MPI)　compensation　in　pulsed　Time-of-Flight　(ToF)　depth　camera　works　well　on　desktop　computer.　However,　pulsed　ToF　depth　cameras　are　mainly　applied　in　fields　such　as　autopilot　vehicles　and　robots.　In　these　situations,　exporting　depth　images　to　a　computer　and　running　MPI　compensation　in　MATLAB　every　time　are　not　realistic.　Therefore,　according　to　the　hardware　structure　of　the　multipath　interference　real-time　compensation　platform,　we　use　QR　decomposition　to　optimize　the　sensing　matrix　in　our　algorithm　for　better　hardware　implementation,　while　maintaining　the　accuracy　of　our　algorithm.　For　the　optimized　algorithm,　we　designed　an　efficient　parallel　data　processing　hardware　structure　that　matches　the　features　of　our　algorithm　to　optimize　the　calculation　efficiency.　Then,　we　implemented　it　on　the　realtime　compensation　platform.　The　experimental　results　show　that　this　algorithm　architecture　runs　stably　under　the　clock　frequency　and　it　runs　faster　than　desktop　computer.　©　Published　under　licence　by　IOP　Publishing　Ltd.