Algorithm 4: Diagnosis_Request_Service

Input: skin_image, service_type Output: class of skin disease

1   Function: skin_diagnosis (skin_image, service_type) 2    Init: ip ← grpc server ip address, crt ← server certificate, url ← cloud service URL, 3       service_id ← nearby service ID, user_name ← given device name 4    Try 5       If service_type = = mobile_local_service Then           //request from the local service 6         P[p0, … ,pC] ← mobile_local_service.get_diagnosis (skin_image) 7       Else If service_type = = mobile_remote_service Then           //Create connection with the nearby device 8           connection ← request_connection(user_name)           //Send a request to the diagnosis service 9           P[p0, … ,pC] ← connection.get_diagnosis (skin_image) 10      Else If service_type = = grpc_service Then             //Create a secure channel with the diagnosis service 11         channel ← create_secure_channel (ip, crt) 12         P[p0, … ,pC] ← channel.get_diagnosis (skin_image) 13      Else If service_type = = containerized_grpc_service Then             //Create a secure channel with the cloud 14         channel ← create_secure_channel (url) 15         P[p0, … ,pC] ← channel.get_diagnosis (skin_image) 16      End If           //Find the largest probability value 17      probability ← p0 18      prediction_class ← 0 19      For c in C 20         If P[c] > probability Then 21             probability ← P[c] 22             prediction_class ← c 23         End If 24      End For           //Find corresponding labels 25      prediction_label ← get_prediction_label(prediction_class) 26      Return prediction_label 27   Catch exception 28      Return error_message 29   End Try