Billions of data is getting connected to Internet of Things server but many a times the data is from different systems, especially in Industries a lot of data comes to same server through different gateways. Some typical applications like Healthcare, Transportation, and Smart City are looking to connect to the Internet of Things server but through a gateway of its own, which increases the complexity in system architecture. Traditional Internet of Things Gateways is used to collect the data from downstream devices connected to sensors and send it to the respective server for further processing. But however there is a need for typical heterogeneous gateways which can minimize the traffic and with presence Machine learning technologies, so the intelligence can be brought into the gateway itself for further prediction and processing. It is predicted that by 2022 the need of Internet of Thing gateways is not only to process the data stream but also apply some intelligence and if possible predict the outcomes based on the requirement, which means the Internet of Things gateway Architectures are going to be quite different from the traditional gateway architectures. Also it is important in cases like Smart Cities where there is a need of reading data from different heterogeneous devices which relies on its own architectures and network protocols like WiFi, ZigBee, LoRaWAN, Sigfox it is understood that if in case of these scenarios if there can be a single gateway which can read data from heterogeneous devices and process the data stream for further processing there can be much better understanding of data processing. This can also minimize network traffic, radiation hazards. The following work emphasizes on the need of Multi-standard gateway and relevant case studies focus on implementation of machine learning algorithm in predicting an appropriate result. © BEIESP.