4 ms·
For those who's not familiar with V2V standard, or think this is useless, here's some more context. The whole standard stack is called Wireless Access in Vehic
by songgao 13y ago
For those who's not familiar with V2V standard, or think this is useless, here's some more context.
The whole standard stack is called Wireless Access in Vehicular Environments (WAVE) / Dedicated Short Range Communication (DSRC). Short range is as in being shorter than cellular networks, which is easily over hundreds of meters.
Physical Layer and Link Layer of WAVE/DSRC are defined in IEEE 802.11p. It's so much like other protocols in IEEE 802.11 suite that it's only an 802.11 amendment and is now integrated into IEEE 802.11 standard. In US regulatory domain, 802.11p uses 5.855 GHz - 5.925 GHz, which is slightly higher (and not overlapped with) other 5 GHz 802.11 protocols like 802.11 a/n/ac, so that V2V won't be interfered by WiFi devices. Above Link Layer is not conventional. It's designed to support TCP/IP on top of that, but just for non-safety messages. Safety messages are supposed to use WAVE Short Message Protocol (WSMP), which replaces IP Layer and Transport Layer in TCP/IP suite. WSMP is believed to be more efficient in V2V context.
A promising application of WAVE/DSRC is Cooperative Adaptive Cruise Control (CACC) and Emergency Braking. Consider following scenario:
---------------------------------
A B C D E
---------------------------------
where ABCDE are five vehicles on highway. In ACC, each vehicle uses radar, infrared, etc. to detect distance between the vehicle in front of it, and adjust speed. Theoretically this may reduce safety distance because machine reacts faster than human being and in case of emergency the vehicle can brake by itself. However, it still cost time to detect the change of distance, to calculate, and to brake. So if A encounters collision, B would brake, then C detects distance change, and brake, then D brakes. When it comes to E, accumulated time is much longer. This means you can't really reduce distance that much if you want to rely on automatic emergency brake. With V2V enabled, each of the vehicle keeps broadcasting vehicle broadcasting so they all know each other's condition. If A encounters a collision or brakes very hard, E would know almost at same time when B knows. This means E can react much faster, so vehicles can be much closer to each other. This is very useful for trucks considering the aerodynamics. Also, with vehicles closer to each other, you can put more vehicles in the same high way.
Other applications include interaction collision warning, travel time prediction, and integration with pedestrian safety system etc. The key thing is V2V is localized, distributed, and thus real-time and more responsive than cellular connection. It's definitely gonna enable more exciting technology advances in automotive industry.
- teddyh 13y agoI doubt vehicles are special enough to require their own replacement for TCP/IP. Last time I heard that kind of talk was with WAP/WML – anyone remember those? TCP/IP is good enough, and making your own protocol on that level is never worth the huge hassle for everyone involved. Unless your aim is to obfuscate and erect barriers to entry.
- songgao 13y agoUpvoted. I agree that it's annoying and confusing. I don't like it either. But that's the standard that IEEE, ITS and USDOT are supporting and car/truck companies are working on that. Unlike WiFi routers, vehicular networking requires everybody to use the same standard in order for it to work. I'm not sure how much space there is to make an effort to build an alternative. They have their reasons for replacing TCP/IP, among which one is to reduce IP overhead in vehicular environment. Since most of time communication is temporary and doesn't require routing, IP layer is not necessary. They want to reduce the IP header overhead for each packet. I doubt how much it improves reliability though, given the fact that people have to implement another whole stack other than just using mature TCP/IP stack.