router upgrade recommendation for distance

[Abe18]

I recently joined this forum and have read a number of the thorough reviews as well as the posts. However, I have not seen anyone specifically address what router would be best in the following situation: We have two home offices in our split-level home. In the basement, a desktop is connected to a Linksys 802.11b wireless router (BEFW11S4 v2) via a CAT-5 cable coming from a 3Com OfficeConnect 10/100 Network Interface Card (3CSOHO100B-TX). On the top floor (1 1/2 floors up), on the opposite side of the house, in a second office is a laptop connected via an external PCMCIA card. The distance from the router to the laptop is about 55-60 feet on a straight line. The laptop accesses the files on the desktop and also uses the desktop as a mail server. Finally, a second printer with wireless (802.11g) capability was recently purchased and will be used by the laptop, but only sporadically. Since I cannot justify the expense of a dual band router, I have been looking at single band n routers. However, I am having difficulty in determining what "n" router is best for usage at 60 feet. I am aware that when the wireless printer is printing, it will slow does the n connection, but I am willing to live with that since the printer will only be used occasionally. Belkin's N+ router appeared to be a good compromise now that its price has come down, but the performance of this router in locations E and F on your chart suggest that it doesn't perform well over long distances. Is this true? If so, what do you recommend? Finally, I intend to replace the current NIC on the desktop with a gigabit NIC, but don't know how to determine which card my desktop (Dell Dimension 8200 [bios A01], running XP Pro sp2) will support. Can you point me in the right direction? Thanks.

 

[thiggins]

Every wireless environment is different and distance is only one factor. All you can do is look at the throughput vs. location charts and the descriptions of the test locations and try to match the locations with their equivalents in your situation.

But you really should use the chart information on a relative basis, not as an absolute measure of performance. Keep in mind also that the client used will also make a difference. Which draft 11 clients do you/ will you have?

As for gigabit NICs, I would be surprised if any gigabit NIC wouldn't work in your Windows machine. The main recommendation is to get a PCIe vs. PCI NIC, if your machine supports PCIe.

 

[Abe18]

Thanks for your response. Is it possible to search your charts filtering for best performance at long distances? I didn't see that that is an option.

What is a draft n client? If you are referring to the software that interacts with the router, wouldn't that come with the router? If it has something to do with the modem, I am using a Linksys Etherfast Cable Modem, model BEFCMU10.

As for PCIe, are you suggesting that this would be preferable to getting a gigabit NIC for the desktop and connect via wireless rather than cable?

Abe18

 

[thiggins]

Thanks for your response. Is it possible to search your charts filtering for best performance at long distances? I didn't see that that is an option.

No. Sorry that is not a filter option right now.

What is a draft n client? If you are referring to the software that interacts with the router, wouldn't that come with the router? If it has something to do with the modem, I am using a Linksys Etherfast Cable Modem, model BEFCMU10.

A "Client" is a device that connects to a wireless router or AP.
To get a throughput advantage from using a draft 11n router your computers need draft 11n adapters either built in or added on.

Adding a draft 11n router will not expand your wireless LAN range. It can provide higher throughput at a given location.

As for PCIe, are you suggesting that this would be preferable to getting a gigabit NIC for the desktop and connect via wireless rather than cable?

PCIe is a bus standard. You can get gigabit Ethernet adapters for PCI, PCIe, Cardbus and even USB 2.0.

 

[Abe18]

Ok. So "client" is nothing other than the adapter card, right? If so, I was fully aware that I needed to have a "matching" adapter card to go along with the N router. Indeed, when I bought the laptop 7 years ago, I expressly ordered one with an external pcmcia slot so I would be able to switch the adapter card when I upgraded my network, as I am doing now. Since my understanding is that one should use a card that is matched with the router, until I know what router I will be purchasing, I don't know what card I will be using. Regarding the Belkin N+ router, it doesn't appear that Belkin makes a pcmcia card expressly designed for that router, rather, Belkin has a USB N+ adapter. Although Belkin does have a pcmcia card which is an N card, I don't know if I would lose speed or reception by using the N card rather than the USB N+ adapter. I would prefer to not use the USB adapter since I need the USB port for other devices.

//Abe18

 

[Abe18]

How to interpret link test results for use with throughput charts

Can I use the information from my wireless client's (Dell TrueMobile 1150) link test to inform me as to what I should be looking for in terms of the data presented on the downlink throughput charts? The link test shows that the signal level is about -65dBm, the noise level is about -90dBm, and the SNR is about 25dB. This has remained constant (plus or minus a few dBm) for the past hour. What I am trying to determine is whether replacement of the 801.11b router with an N router and matching client will result in not only a faster throughput, but also an improved signal. In what way, if any does the Path Loss data in the Azimuth charts relate to any of the data produced in my wireless client's link test? It appears that the D-Link DIR 655 "falls off" at about 103dB, which appears to be better than most routers. However, is there anyway to use this information comparatively with the data from my wireless client's link test?

Abe18

 

[thiggins]

No. The information presented by the wireless utility is not calibrated to a fixed standard. And signal level does not accurately predict throughput. There are too many variables.