How it works:
PLM NETWORK PROTOCOL
In our intelligent street light application, the network is composed of several power line modems, called nodes. These modem nodes are distributed throughout the lampposts, each of which holds the lights with their power supply. One of the modem nodes is used as a data concentrator, normally located in the power cabinet (which distributes the power to a certain group of lampposts) operating in a three phase power line. Each device can be connected with up to three different power phases. In this way, all the nodes and the concentrator share the same power line, which is used as a data communication channel (physical means).
In this application, the concentrator is controlled remotely via GPRS modem by a Remote Service Center (RSC). Here all the information about the lamps - consumption, status, and errors - as well as information about the cabinet itself - (ambient temperature and breakers status) - are relayed to and stored in a database. A web server with a dedicated interface allows the RSC to analyze the data and perform many remote actions: changing the remote lighting schedule of on/off/dimming, manually switching on, off or dimming the lamps, or changing the internal time clock of the modems.
The network is logically designed to be a master-slave structure, where the data concentrator is considered a master device and each node is the slave. And as a matter of fact, any device can initiate the communication, thereby becoming a master, while the target node, each having a unique address ID, becomes the slave device.
Each node can also act as a data repeater, without any additional programming features, increasing both the reliability of the network and the statistical probability that the information from the master to the slave will be delivered even under difficult network conditions.
However, the coexistence of more than one master and more repeaters introduces the need of the data collision management. Since more than one device can initiate communication at any time, this could cause a network jam, lowering overall performance. This potential problem can be solved using one of two main techniques. One of these - carrier sense multiple access with collision detect (CSMA/CD) - is used when a hardware device is able to detect any collision during the data transmission. The other - carrier sense multiple access with collision avoidance (CSMA/CA), which is the one we will describe in this article – is used when the hardware doesn’t have this capability.
The implemented conflict avoidance mechanism uses a back-off time and the “band in use” (BU) hardware feature of the PLM to avoid transmission conflicts. Before initiating any communication, each device waits until the band is free by checking its own BU flag. As soon as the band is free, a random back-off time is calculated. Once the back-off time is elapsed, if the band is still free, the transmission is started, otherwise the loop is started again (waiting of the BU and new back-off calculation).
The data exchange between all the nodes connected to the same power line also uses a data frame acknowledgement mechanism. In this way, the master knows if its own transmitted data packed has been correctly delivered to the target node since it receives an acknowledgment frame for each data frame sent to a target device. This acknowledgement excludes some data frames sent in broadcast by the master.
The repetition feature improves the probability that the data frame will be delivered to nodes that are far away or under noisy network environments. As all the other nodes are connected in the same power line, they continuously sense the network catching all data transiting on it. Depending on the device addressing and the data frame/acknowledge flow, each node is able to detect if the sensed frame has to be repeated, discharged, or processed. A data identification technique (frame ID) and the forward error correction redundancy (FEC) in the data frames are implemented in order to avoid cyclic repetitions or data loss that can cause an unnecessary increase of data traffic.