Blockchain-based　Industrial　Internet-of-Things　(IIoT)　integrates　the　blockchain　technology　into　the　traditional　IIoT　infrastructure　to　provide　secure　and　collaborative　services.　In　IIoT,　the　traffic　is　usually　encrypted　using　cipher　suite　(SSL/TLS)　for　secure　communication,　which　makes　it　hard　for　middleboxes　to　detect　malicious　activity　in　the　traffic.　To　address　this　problem,　secure　middleboxes　that　directly　perform　encrypted　traffic　inspection　have　been　presented.　Recently,　a　new　privacy-preserving　deep　packet　inspection　(DPI)　system　on　middlebox　for　IoT　scenarios　was　proposed,　but　it　suffered　from　the　following　two　limitations:　(i)　no　support　for　fast　token　detection;　(ii)　no　support　for　tracing　abnormal　sources.　To　address　the　two　limitations,　we　propose　BTDPI,　a　privacy-preserving　traceable　DPI　system　that　efficiently　performs　inspection　over　encrypted　traffic　in　Blockchain-based　IIoT.　Technically,　we　adopt　a　two-layer　filter　architecture　to　improve　the　efficiency　of　detection,　and　moreover　introduce　a　new　online-offline　certificateless　aggregate　signature　with　smart　contract　to　design　an　identity　traceability　mechanism.　The　experiment　result　shows　that　BTDPI　runs　26.7×　faster　for　token　detection　with　3,000　tokens　and　3,000　rules　than　the　state-of-the-art　work.　IEEE