Hi all, I would just like to make a very short report on the meeting that took place yesterday with the WG chairs of ROLL - Routing Over Low power and Lossy networks.
>From ROLL : JP Vasseur and David Culler >From TICTOC : Yaakov and Laurent (Stewart was held up in the routing area >meeting) The point of this meeting was to fill in the matrix column for distributed sensor networks. For those who have not been following ROLL work, it is concerned with networks composed of many embedded devices with limited power, memory, and processing resources. These devices are interconnected by wireless links, such as low-power WiFi, bluetooth, or IEEE 802.15.4. In order to conserve power each device sleeps most of the time, and exchanges packets only with nearby devices. The devices can only contain inexpensive 15ppm crystals, and even these are turned off most of the time. Since the sensors only communicate with neighbors, the routing is geometry dependent, and interacts with timing. There are already two time distribution protocols in this space FTSP - Flooding Time Synchronization Protocol (available in open source) TSMP - Time Synchronized Mesh Protocol (proprietary protocol - no full description available) There are two approaches to handle the interactions between routing and timing. In one approach timing information is piggybacked on the existing routing protocol, and thus benefits from the authentication, encryption, etc. In the other approach the timing is considered more fundamental, and the routing information is piggybacked over the timing protocol. >From the timing point of view there are three different applications with >different levels of required performance: 1) instrumentation / measurement: In this application there are mulitple sensors (e.g. temperature sensors) placed around a building The time accuracies required are on the order of tens of millisec. 2) structural monitoring Here multiple sensors (e.g. position or motion sensors) are placed on a structure, and it is required to measure modes of stress or motion. The time accuracies are sub millisec. 3) TDMA / frequency hoping Here in addition to the sensor waking up at a specific time, the transmission is frequency agile. The time accuracy required is 10s of microsec. David agreed to help fill in our matrix for these three cases. Y(J)S _______________________________________________ TICTOC mailing list [email protected] https://www.ietf.org/mailman/listinfo/tictoc
