LAN vs WiFi

[KishSquared]

Hey all,

I just purchased a LAN adapter, and I'd really like to run some tests to see if it's truly better than WiFi for playing Brawl online. I don't expect to see a difference unless I play with someone that also has a LAN adapter. Are there any lab members that have a LAN adapter?

Just an FYI - I'm a wireless network engineer by trade (sporting multiple Cisco certifications to back it up!), so I'm intimately familiar with the technologies behind LAN and WiFi. I'm well aware that the speed of WiFi is plenty good for playing a game like Smash, but interference can cause degradation. I have personally seen this twice - once from direct interference (from a Wavebird, no less), and once from environmental interference caused by other WiFi devices. Therefore, it's no debate that WiFi in inferior to LAN, the question is simply by how much, and whether it's worth it for people to purchase these devices.

So in short, I'm not looking for a debate or discussion on what should or should not be the case with WiFi vs LAN. I'm looking for some hard data by actual experimentation. So... any takers?

 

[infomon]

Is there any published info about the network topology Nintendo employs for its wi-fi stuff? I mean, in a 1v1 match is data being routed through some nintendo servers or does it actually do a "direct" internet connection, the hosting wii acting as a server? and etc. miscellaneous other questions about what sources of lag exist in wifi matches.........

Actually it would be interesting to try and use homebrew stuff to modify the WiFi engine so we could do WiFi matches organized through non-Nintendo servers. It might be as simple as changing an IP address somewhere. Or if it's using DNS to resolve nintendo's wifi server, you could prolly just spoof that within the context of a LAN to experiment. Hmmmmmmmm.

 

[SamuraiPanda]

A long time ago I remember some network-savvy individuals mentioning that WiFi and LAN are equivalent. And I also remember them saying that all online is routed through Nintendo servers.

 

[Levitas]

I'm curious. How would you go about testing this?

 

[ColinJF]

infzy & SamuraiPanda: It's just peer to peer. I've checked out the logs on my router (I have dd-wrt) while using WiFi and it's direct UDP/IP (i.e. you are sending UDP packets directly to the opponent's IP). (You can even use this to learn your opponent's IP address.) The only thing Nintendo servers are used for is the lobby system (which is done using TCP/IP by the way). When you join a game, Nintendo communicates to you the opponent's IP, and it's just peer to peer from there. There might be an alternative infrastructure in place if it can't get the peer to peer to work, but I haven't personally seen any evidence of this.

(Note: The type of "peer to peer" we are talking about here is similar to how you download a torrent. Even though one peer probably performs server like tasks, it isn't a client-server relation like you have with tcp.)

As for the LAN adapter as far as I can tell the only difference it should make is it would take a lot less time for packets to go between your Wii and router, but that time is already pretty small (< 3 ms, which is nothing compared to the total latency). There could be some other effects though; I haven't really investigated.

 

[KishSquared]

Well, like I said, I'm a wireless network engineer by trade and I'm well aware that wireless vs wired in a perfect environment are not going to be any different. But wireless interference causes lag, I've seen this on the two occasions I mentioned. So my curiosity is in regards to how much lag is caused by people having wireless interference.

Frankly, in all my experiences, I was shocked to find that interference from other WiFi devices caused lag in Smash as bad as it does. I play plenty of online games via laptop, and never have I personally seen lag be so bad that the channel on the wireless router needs to be changed. That happened to me many months ago, where Brawl was lagging pretty bad online. I ran a channel scan and found that a lot of my neighbors had the same channel, and so I switched channels. The difference was astounding.

For testing purposes, it would be as simple as playing a few matches via WiFi, then plugging in the LAN adapters and playing it that way. I could record the matches and see if there's really much of a difference. It wouldn't be any more complicated than that. I'd sure like to find people from multiple regions to test with as well, log router hops via a tracelog and see how much of an effect that has. Eliminating wireless from the equation will really tell us a lot.

 

[KishSquared]

Chibo and I played about 10 matches on LAN. I hosted every match, and Chibo and I both noticed minor input lag. Chibo noticed it a bit more than me, but I hosted and I'm far more used to online play, so it's hard to tell whether it was any different. There was zero visual lag, and neither of us noticed any missed inputs (meaning that we hit the button and nothing happened).

Switching to WiFi, visual lag and missed button inputs were immediately noticeable. Chibo hosted and said that he thought that the input lag may have been better than on LAN, but being the host might have affected that. Either way, it was very clearly worse than with LAN adapters.

I ran a traceroute to Chibo's IP, and he was about 15 hops away (WI -> NJ). I had a 55-60ms response time when I pinged him.

On a sidenote, my wife started streaming NetFlix partway through a LAN match, and it produced very noticeable lag. In other words, downloads greatly affect performance on Smash matches.

Good stuff, Chibo! You rock for doing that with me.