Through the use of billions of data points, the internet of things (IOT) links billions of objects to the internet, all of which require security. A major issue for cyber security is the expanded attack surface of IOT. To overcome these challenges, Intrusion Detection Architecture (IDA) has been proposed in this paper, which helps to trace the Intrusion Data in IOT. The proposed Intrusion Detection Architecture (IDA) is performed in three stages namely, Data Collection, Pre-processing in Data encoding and Classification block. Initially, the request from the IOT devices is sent to the Data collection (DC) and pre-processing stage there it could find the amount of data. The collected data traces and filter through the Normalization Technique (NT). Then the filtered data goes to the data encoding block. After the encoded data goes to classification block here it classifies the data by Ghost Net technique and finally the attack can be classified and detected. The effectiveness of the suggested IDA strategy has been assessed using assessment measures such as configuration latency, detection rate, accuracy, precision, recall rate, false detection rate. The proposed method reduces the communication overhead of 75%, 50% and 38% than SSTS, ANT and FPDS existing techniques.