I noticed that using BitTorrent made my internet connection effectively useless for other things: web browsing, video chat, VOIP, etc. To fix this, I installed Tomato by Shibby firmware on my Linksys E2500 router and turned on the QoS (Quality of Service) settings. Here's how I configured QoS.
The first step in setting your QoS settings is to determine your internet speed. I used the tests at TestMy.net: Upload Test / Download Test. I got the results below:
Next, I started downloading a bunch of legal torrents to try to max out my internet connection. I used Ubuntu torrents for this. This gave me 14 several-gigabyte torrents with a large number of seeders. My internet connection was quickly saturated. I did another speed test, and the results show how badly bogged down my internet connection was:
In order to tell the router how to prioritize your internet traffic, you first have to tell it how to classify each type of traffic on your network. I used the Wi-Fi Planet settings with a few modifications (under QoS > Classification):
I kept my QoS Basic settings as simple as possible. Probably the most important part of these settings at this point is that the default class is "3-Medium". This class is prioritized over BitTorrent because it could contain other things like VOIP, but it's not much higher than BitTorrent because BitTorrent is hard to classify. Any BitTorrent connections that leak out will get lumped into this category too.
Finally, I had to fine-tune the Inbound and Outbound Max Bandwidth Limits. I did this in a relatively straight-forward manner. While my connection was saturated with the Ubunutu torrents, I set the limits to different values and re-ran the upload and download tests. My results are below:
I highlighted the bandwidth limits that gave me the best results. Each setting that I tested was 5% lower than the previous setting, and the maximum setting was 100% of the upload/download speeds I got with an idle network.
There's an important trade-off that should be understood here. If you set the bandwidth limits too high, then QoS will never step in and throttle back your torrents to make room for web browsing. This will result in really bad web browsing performance when you have torrents or downloads running. On the other hand, if you set the bandwidth limits too low, then QoS will be throttling all of your traffic all of the time. This will result in slower torrents, slower downloads, and slower web browsing.
Photo by Torkild Retvedt
I've got an old router. I've had it forever. I know I definitely had it my Freshman year of college back in 2007, but I don't know exactly when I got it.
Anyway, I've recently been having some problems with the Wi-Fi in my apartment, so I ventured into my router's settings to try to see if I could fix it. I noticed that the date and time were set incorrectly. When I tried to set the time, I found out my router doesn't believe in life after 2012. Or at least the people that built my router didn't expect it to continue functioning past 2012:
That's right. The date dropdown in my router's settings page only goes up to 2012. I think I won't buy any more D-Link routers now.
What might be more surprising than that is that there is a vendor on Newegg selling the exact same model router as the one that I have for $79.48! Look at that overpriced relic.
Linksys has a good reputation, and a lot of their routers support DD-WRT, so I gave them a look. I found a refurbished Linksys N600 / E2500 on Amazon for under $40, so I scooped it up.
Linksys N600 / E2500
It's moderately priced, has 2 antennae, and supports both the 2.4 GHz and 5 GHz bands. And it runs DD-WRT!
I actually used the setup CD for the E2500. That's an unusual thing for me because I'm suspicious of most software provided with any electronics I buy because it's usually not very good. The setup for this router was actually very useful. It got everything up and running in under 20 minutes (without using any ethernet cables!) and at the end it gave me an opportunity to configure advanced settings too. Good job, Linksys. Your setup tool was useful.
All of the devices in my home connected extremely quickly to the new router, and I haven't had any connection issues with it at all. I'm really happy with this new router. I might not end up running DD-WRT on it after all.
Photo by Andrew Hart
My desktop recently died on me, so I'm looking to build a new computer. Part of the build includes a USB Wi-Fi adapter. The last Wi-Fi adapter I bought was 802.11g. Since then, 802.11n and 802.11ac have come out and wireless networking has gotten a lot more complicated.
A user reported a connection speed of only 65 Mbps between his laptop and router even though both were advertised to operate at up to 150 Mbps. Why? His laptop can operate a pair of simultaneous 20 MHz streams at 72.2 Mbps each. His router, on the other hand, supports only one stream but with a 40 MHz bandwidth.
This combination is limited to one stream by his router and a maximum channel width of 20 MHz by his laptop. The fastest speed possible with a single 20 MHz channel is 72.2 Mbps when using a short guard interval. Since his setup is running with a long guard interval, the speed is reduced to 65 Mbps.
The take-away is to make sure that you match the number of streams and channel widths between your devices for optimal speed. That doesn't sound too hard, but it is. Manufacturers don't readily advertise that information. Instead they use wonky class designations that are ambiguous and sometimes misleading.
I've bolded the class designations below that I recommend. Products with these class designations are most likely to be 100% compatible with your existing devices and are the least likely to feature proprietary extensions to the 802.11 standards that won't work with devices from other manufacturers. Specifically, these class designations are N600, N900, AC1200, and AC1750. (The N300 designation is generally also free of proprietary extensions, but higher performing N600 routers can be purchased for approximately the same price.)
Decoding the wireless N class designation is pretty simple. See the table below for how D-Link explains wireless N class designations.
Basically, take the number after N, divide by 300, and round down to get the minimum number of streams you have on a given frequency band. (For N750 we have 750/300 = 2.5 → at least 2 streams on the 2.4 GHz and 5 GHz bands.)
If your device is not dual-band, it only operates at 2.4 GHZ or 5 GHz but not both. In this case, use 150 instead of 300 in the calculations above. (For single-band N450 we have 450/150 = 3 → 3 streams.)
These calculations are assuming a 40 MHz channel. This is not always the case as evidenced by the issue at the beginning of this post. For more information, see the chart of 802.11n data rates.
Decoding the meaning of the wireless AC class designations is a bit more difficult. The table below gives my best guess for each term using the standard 802.11ac data rates. Beware that different combinations could be used by manufacturers to get to the same class designation. Always read your device's specifications carefully.
*Proprietary extensions to the 802.11n specification.
**Not widely available as of September, 2014.
Photo by Sean MacEntee
There are a number of misconceptions about 2.4 GHz and 5 GHz Wi-Fi. Let's clear those up.
In the US, the FCC set aside the frequencies 2,400-2,483.5 MHz and 5,150-5,725 MHz for unlicensed broadcasts such as Wi-Fi, cordless phones, etc.* These regions are known as the 2.4 GHz and 5GHz bands, respectively.
The upper frequency of a band minus the lower frequency of a band is the "bandwidth". The bandwidth of a frequency band is directly proportional to the rate at which data can be transmitted. The bandwidth of the 2.4 GHz band is about 80 MHz and the bandwidth of the 5 GHz band is about 570 MHz. Therefore Wi-Fi can transmit data faster on the 5 GHz band because it is wider.
*The ranges given here aren't 100% accurate. The way the FCC splits up the electromagnetic spectrum is complicated.
Microwaves are a huge source of interference for Wi-Fi. Microwaves heat up food by showering it with 2.4 GHz radiation. Some of this radiation leaks out and confuses Wi-Fi receivers when the microwave is running. Cordless phones also often operate on the 2.4 GHz band. The 5 GHz band is used by fewer devices, so the amount of interference on it is less.
The radio engineer formula gives the path loss in db:
where L is the path loss in decibels, d is the distance traveled by the signal, f is the frequency of the signal, and c is the speed of light. Based on this equation, we can expect the path loss to be 20*log10(5/2.4) = 6.3752 dB more for a 5.0 GHZ signal than a 2.4 GHz.
Transmit a pair of signals at 2.4 GHz and 5.0 GHz with 100 mW of transmit power each. If the 2.4 GHz signal received is 10 mW, then the 5 GHz signal will only be 2.304 mW at the same receiver.
Photo by Bill Smith