Archive

Posts Tagged ‘Wifi’

Mesh Networking (or 802.11s) comes to Linux

September 4, 2008 Leave a comment

Since kernel 2.6.22 there has been a huge drive to rework and update the entire wireless networking stack in Linux. Back in July, 2.6.26 was released with over 600 wifi specific changes including :

  • mac80211 now supports draft mesh networking (802.11s; thanks to the open80211s project
  • mac80211 now supports more optional HT (802.11n) features
  • mac80211′s monitor interfaces can now be configured more precisely, “cooked” monitors were added
  • mac80211′s IBSS implementation now supports IBSS merging and received various bugfixes
  • mac80211 now has an in-kernel documentation book, help welcome!
  • b43 now has support for PCMCIA devices and QoS/WME

The following have been removed.

  • the bcm43xx driver
  • the old ieee80211softmac code

A while ago I discussed the deprecation of the bcm43xx driver so I welcome its official removal from the kernel as this means less time wasted disabling this legacy kernel module prior to installing / activating b43 on every new install. The mesh networking (draft specification designated 802.11s) support is largly thanks to the open80211s project however it is only in the fairly early stages. Support is not consistent across all the Linux wifi drivers but full support is progressing. I do wonder whether mesh networking will take off, it is however, an exciting concept. Imagine the following scenarios:

1. A dense build up urban area, where multiple nodes (flats / homes etc) could connect via one or two Internet connected nodes.

2. A neighbourhood which contained a lot of like minded PC users each connecting as a node in a local mesh network to create an ‘instant’ LAN/WAN or gaming network. Imagine being about to connect Xbox 360s / PS3 / PSP or even PC gamers wirelessly to your neighbours without using the Internet services.

3. A sparse, less developed environment where multiple nodes are greatly spread out and Internet access is rare.

The possibilities are endless, however with all such technologies, a lot of people need to jump on the bandwagon, particularly with ad-hoc technologies such as these.

Windows Media Center 2005 woes

July 23, 2008 Leave a comment

I managed to acquire, for the price of a nice lunch, a brand new Elonex media center Artisan LX a couple of days back. I was initially very excited because up to then I had still been running my first media center was really just an experiment, built from scratch containing mostly old components I had around my place. A year and a bit on, I am firmly hooked on a PC based PVR system the cornerstone of my entertainment system. It contained an Athlon 2600+ processor with 512Mb of DDR coupled with a DVB-T Hauppage tuner and an 80Gb drive for recordings running the open source MediaPortal software. So as you can see, there was plenty of room for improvement.

This was the first time I have really had a tinker with the Windows Media Center range of Operating systems that Microsoft produce and I went in with few expectations, apart from wanting at least as comparable an experience in terms of functionality and flexibility as I have enjoyed with MediaPortal.

The first thing that struck me was how fickle Windows Media Center 2005 was, even with all the roll ups (essentially what Microsoft call Service Packs for Media Center OS) installed. Wikipedia sums up the ‘capabilities’ of WMC 2005:

‘Media Center originally had a limitation of 1 analog tuner, but was raised to 2 analog tuners with Media Center 2005. With Update Rollup 1 for Media Center 2005, support for a digital tuner was added, but an analog tuner must still be present for the digital tuner to function. With Rollup 2, up to 4 TV tuners can be configured (2 analog and 2 HDTV). All the tuners must use the same source, for example they must all be off an aerial or a set-top box using the same guide data, you cannot mix Sky Digital and DVB-T for example.’

XP Media Center really shows its age here – I do not watch any analogue transmissions, so for a Media Center to require a legacy piece of hardware just to be able to access DVB (digital) seems preposterous. But that was not the worst thing! Windows Media Center 2005 is not capable of pulling EPG data OTA (over-the-air) instead requiring an overly elaborate system that relies on a permanent, always on Internet connection. This also raises some privacy concerns as ‘anonymous’ data, which is not entirely anonymous as Microsoft asks for your postcode during set up, is fed back to Microsoft which can include recording / watching trends and general EPG usage. Hitherto my media center system has not been networked. Considering it is in the opposite corner of my house, and I do not stream my recordings or have formal media shares, I never felt the need to network it. It was nice to just have a static, secure system without any security programs or periodic updates – now security monitoring of my media center has been added to my list of digital chores.

None the less, I was determined to give it a fair go, so I added a wifi adaptor, added some plug-ins and configured everything. After spending eight hours getting everything working, playing around and testing… I went back to my custom build. Not all the problems can be put squarely at Microsoft’s feet however. Elonex declared bankruptcy shortly after launching this range and the malicious part of me can see why, if this mediacenter is the sum total of their expertise.

Whilst the case looked rather nice from the outside, the hardware and the design of the internals is what really lets it down. The only element Elonex got right was the noise (or lack thereof) – the media center barely gives out a murmur when idle due to only a since fan which is housed inside the power supply. It runs at 690rpm, which draws air over the CPU heatsink (which has four heat pipes) and directly out the side of the case. However, I stressed ‘at idle’ before for a reason. When the media center does anything the incredibly noisy hard drive starts very audibly clicking and crunching away and it completely lets the machine down.

However that’s not the worst thing about this mediacenter. Due to the fact that there is only one very slow fan the airflow in the case is restricted to circulating around the motherboard tray, the processor then out the power supply. The harddrive and PCI / AGP cards are completely neglected. This point was slammed home when the harddrive consistently reported temperatures of high 50s to 62 degrees Celsius!!! Worse still, when I idled the system, that heat didn’t dissipate. The hard drive is locked into place with a pretentious plastic locking mechanism which neither improves the accessibility of the drive bay nor decreases the vibrations from the drive. There is no thermal (or thermally viable) contact between the hard drive and the case and as such, the hard drive is left smouldering away with no way to cool down predictably with next to no drop in temperature. There is a valid point that maintaining electronic components at a set temperature prolongs their life by avoiding constantly repeating thermal differentials (i.e. heating and cooling) however the fact remains that 60+ degrees centigrade is far too hot for a hard drive. Although my brief research on this did not yield any definitive threshold, most sources agree that 50-55 degrees Centigrade is about the absolute maximum recommended operating temperature.

Couple this practically zero thermal conduction with a lack of airflow and you have a recipie for a very short hard drive life. Even worse, this thermal issue was not limited to HDD, the south-bridge and GFX heatsinks were equally poorly cooled and get unpleasantly hot to the touch.

Worst of all, it is just slow. CpuID and the BIOS disagreed with each other about the exact Intel processor that powers the system. I believe it to be either an Intel Pentium 4 530 (at 3.06Ghz) or a Celeron D 345. There is no way the much older Athlon 2600+ processor with the same RAM should be out performing this setup and yet it does so without breaking a sweat.

All in all, very disappointing. A remarkable demonstration of technical ignorance on the part of Elonex. But hey, I didn’t pay for it and now I have an extra DVB-T tuner back in my original, self built machine.

Design (cosmetic) : 8/10 - Pleasing, with a nice Hi-fi look.

Design (technical) : 2/10 – Poor components poorly arranged.

Cooling : 6/10 - Great CPU and powersupply cooling, but everything else is woefully neglected.

Acoustics : 6/10 - Silent until it has to touch the harddrive, still a good effort though

Connectivity : 8/10 – Lots of connectors for digital Audio and Video

Capacity : 5/10 - 200Gb harddrive with a portion taken for recovery. I wouldn’t trust it though and by modern standards it is rather anemic.

Overall : 2/10 – Great for free, if I paid anything for it I would have been annoyed.

bcm43xx is now deprecated (Woohoo!)

December 17, 2007 2 comments

Although the celebration could potentially end there. Every few months I google my laptop’s wireless card (a Broadcom 4318) to see how much further (sic) native linux support has progressed. This is a time honoured ritual I began when I first started taking an interest in running Linux as my main OS on my laptop a few years ago.

Broadcom are a company who seem to care very little about supporting their end users. I have never had to deal with them directly, however and I am basing this statement on their website which seems far more geared to ODMs and OEMs rather than the end user and the fact that they have point blank refused to release source or any useful technical documentation to allow the F/OSS community to make a native driver.

The problem with Google based searches around this problem is that they generally provide links to pages based on Google’s page ranking / relevancy algorithms which do not (for the purposes of this discussion) take page age into account. This means the majority of the results are either wrong or woefully out of date. Up until recently, the main way in which the cards could be made to (sort of) work was through the bcm43xx project. This utilised a reverse engineered framework which required the firmware from the windows based drivers. This was achieved through a sister project – bcm43xx-fwcutter.

While this supported a healthy number of cards, the 4311 still had power issues which would make it very unstable to the point where a hard reset would be required after 15-35 minutes use in order to re-enable the wifi card. So today I began searching again to see what progress had been made and was pleasantly surprised to find that, as the title suggests, bcm43xx the cumbersome behemoth has been depreciated. However the reason I said not to celebrate yet was because it has been replaced the the b43 project. Proudly displayed on the main page under supported devices is the 4318! This hopefully now means that I and other Fujitsu Siemens a1650 owners can now use their laptops as portable devices whilst using a F/OSS operating system!

working

* Station mode
* Access Point mode (although not tested very well).
* Ad-Hoc (IBSS) mode
* Monitor and Promisc mode.
* “Monitor while operating” and multiple monitor interfaces.
* In-Hardware traffic de/encryption (relieves your CPU).
* LEDs to signal card state and traffic.
* In-Hardware MAC address filter.
* Probably something we forgot to add here.

not working yet

* Interference mitigation.
* Bluetooth coexistance (most code implemented, but untested)
* Probably something else that’s not listed under “Works”.

The manner in which it works however is the same, a basic framework is installed with b43 and coupled with the corresponding Broadcom proprietary driver;

“The Broadcom wireless chip needs software, called “firmware” … (T)his firmware is copyrighted by Broadcom … and it must be extracted from Broadcom’s proprietary drivers … you must download the driver from a legal distribution point … (then) you must extract the firmware from that Broadcom driver and install it in the special directory for firmware – usually /lib/firmware.”

Not only has the project been given a new lease of life, the new code is reported to be more stable and resolve the power issues as experience on the Fujistu A1650. It should be noted, however that Acer_Acpi will still be required on this laptop to initialise the hardware but this is beyond the scope of this post.

I will be moving from my “build-and-held-together-by-ducttape-and-a-stern-look” Ubuntu 7.04 build to either 7.10 or Kubuntu 7.10 soon and I shall test this and report back my findings.

Follow

Get every new post delivered to your Inbox.

%d bloggers like this: