I had exhibit the IoT proof of concept at Embedded Technology 2015 in Macnica showcase(Thanks for Macnia all stuff!!). In this exhibit, realtime monitoring, analyzing the sensor data in cloud system by using several sensors and networking protocol.
I had some problem to be solved when I developed. The sensor has it’s original networking protocol(exp: MQTT, REST, TCP, JSON…etc.) difficult to update that all firmware, hard to use the same networking protocol. For example, if all sensors can use MQTT, this problem is piece of cake. But sensor have own original networking protocol. So I made sensor networking exchange GW in Cloud server side. This GW can manage several networking protocols quickly and easily when new sensor join the system.
And next, how to logging and aggregation the sensor data in system and can use explore, visualize, discover data. Of course logging data can be easy to process and routing in system, have to send to some analyzing system like a elastic/kibana, splunk and more analyzing system(Hadoop, Spark). For resolved this problem, I choice fluentd for this logging system. Fluentd have good performance, message routing architecture, many useful plugins, and can be processing sensor data very easily.
And this IoT PoC system scale for many sensors, have some security requirements.

Architecture Overview.

Exhibit in showcase(written in japanese).
Mpression IoT PoC(Proof of concept) held in ET/IoT 2015 – readltime visualizing the environments sensor data from the company offices around the world., Monitoring

– Sensor Devices
EH Terminal
Kibo(CP-1)
Uzuki Konashi
Broadcom WICED
TI SensorTag

Broadcom WICED and TI SensorTag communicate to the Raspberry PI2(BLE), this how to install environment and nodejs source code is here(but japanese).

– Server Side
Cloud : Konoha
OS : Ubuntu 14.04
Sensor GW : golang/mosquitto
elasticsearch 2.0.0
kibana 4.3
Fluentd (td-agent 2.2.1)
Splunk 6.3

sensor devices place in Germany, Hongkong, Shanghai, Taiwan, Dallas, Japan(ShinYokohama, Osaka, Nagoya, Matsumoto, Shibuay, in exhibit showcase)…etc, realtime visualizing sensor data(temperature, humidity, lux) in GoogleMap.

sensor data detail in elastic/kibana.

when the lux parameter become up(dark to bright), lamp image will change.

sensor data summary in splunk.