3 ms·
Lack of effective time signals could hamper IoT development
- niche 12y agoThis calls for a widely accepted "now" specification
- vvanders 12y agoSomeone needs to tell them about dead-reckoning. These are things that have been solved in game dev for quite a while.
- Animats 12y agoThe actual paper: [1] One of the strengths of the Internet is that it doesn't need tight time synchronization to work. Analog TV did; all the network cameras and videotape machines were synched to a time source in New York. Early digital telephony did; all the time-division multiplexed lines were synched to a time source in New Jersey. As for real time, getting a response consistently within N milliseconds doesn't require global time synchronization. A local clock is fine. Real time operating systems such as QNX and VxWorks routinely do that. (The usual test: hook up a square wave generator to a digital input. Have it trigger an interrupt which activates a process. Have the process turn on a digital output. Connect the digital input and output to a scope. Are there any outliers, output signals that are delayed from the expected response time? There shouldn't be, even if there are low-priority background jobs running. Linux can't do that, but a true real time OS can.) There's a demand for microsecond level clock synchronization in the financial community, to measure the delays and record the order in high-frequency trading transactions. And Google apparently uses tight time synchronization for distributed database commit synchronization. But there's no reason to require that outside single industry boundaries. [1] http://nvlpubs.nist.gov/nistpubs/TechnicalNotes/NIST.TN.1867.pdf http://nvlpubs.nist.gov/nistpubs/TechnicalNotes/NIST.TN.1867...
- RIMR 12y agoCouldn't we just piggy-back off of GPS signals? They are, after all, an array of orbiting atomic clocks...