Remote UNLOCK - HTC IMEI (1 PHONE) it's a special service that will helps you to unlock your Qtek, SPV, MDA, XDA or Imate withtout having any hardware, cables or konowledge at your home. All you will make at your home fast and easy.you need only IMEI (can be found on back of your phone) Supported phones: Bluebird: Intermec PB10 Wallaby: Qtek 1010 / Qtek 1020 / Dopod 686 / O2 Xda / Siemens SX56 / T-Mobile MDA / Telefonica TSM400 Falcon: UTStarcom PPC5050 Himalaya: Qtek 2020 / Dopod 696 / Dopod 699 / O2 Xda II / T-Mobile MDA II / i-mate Pocket PC Phone Edition / Orange SPV M1000 / Vodafone VPA / Telefonica TSM500 / Krome Navigator F1 Alpine: Qtek 2020i / Dopod 696i / O2 Xda IIi / i-mate PDA2 / Orange SPV M2500 BlueAngel: Qtek 9090 / Dopod 700 / O2 Xda IIs / T-Mobile MDA III / i-mate PDA2k / Orange SPV M2000 / E-Plus PDA III / Siemens SX66 / Tata Indicom Ego Gemini: SFR v1620 / Vodafone v1620 / Vodafone VPA III Harrier: i-mate PDA2k EVDO / UTStarcom PPC6600 / UTStarcom & Verizon VX6600 / UTStarcom & Verizon XV6600WOC / UTStarcom & Sprint PPC6601KIT / Telecom New Zealand Harrier / Daxian CU928 Apache: UTStarcom PPC6700 / UTStarcom & Verizon XV6700 / Telecom New Zealand Apache Magician: Qtek S100 / Dopod 818 / O2 Xda Mini / T-Mobile MDA Compact / i-mate JAM / Vodafone VPA Compact Refresh: Qtek S110 / Dopod 828 / Dopod 828+ / i-mate JAM (Limited Charcoal Edition) / Orange SPV M500 / Krome Spy Prophet: Qtek S200 / Dopod 818 Pro / Dopod 830 / Swisscom XPA S200 / O2 Xda Neo / i-mate JAMin / Orange SPV M600 / Vodafone VPA Compact S / T-Com TC 500 Charmer: T-Mobile MDA Compact II Muse: Qtek S300 Galaxy: Qtek G100 / Dopod P100 / i-mate PDA-N / Pharos Traveler GPS 525 / CyberBank POZ G300 Wave: HTC P3000 / Dopod E806C / Dopod 818C Vogue: HTC P3050 / HTC Touch CDMA / Alltel & HTC Touch / Bell & HTC Touch / Sprint & HTC Touch / Telus & HTC Touch / UTStarcom MP6900 / Verizon Touch XV6900 / Okta Touch Artemis: HTC P3300 / HTC P3301 / Qtek G200 / Dopod P800 / Dopod P800W / Orange SPV M650 / T-Mobile MDA Compact III / O2 Xda Orbit Love: HTC P3340 / HTC P3350 / Dopod M700 / Dopod D802 / Dopod D805 Gene: HTC P3400 / HTC P3401 / Dopod D600 Elf: HTC P3450 / HTC Touch / HTC Ted Baker Needle / Orange & HTC Touch P3450 / Dopod S1 / T-Mobile MDA Touch / O2 Xda Nova
Elfin: HTC P3452 / HTC Touch (Enhanced Version) / HTC Touch Color / Rogers & HTC Touch / Dopod S1 (Enhanced Version) / T-Mobile MDA Touch 256 Pharos: HTC P3470 / Dopod P660 Trinity: HTC P3600 / HTC P3600i / Dopod D810 / Dopod CHT9100 / Dopod CHT9110 / Orange SPV M700 / Vodafone VPA Compact GPS / SFR S300+ / Swisscom XPA v1510 Polaris: HTC P3650 / HTC Touch Cruise / SFR & HTC Touch Cruise / Dopod P860 / O2 Xda Orbit 2 Titan: HTC P4000 / Sprint & HTC Mogul / Qwest & HTC Mogul / Alltel & HTC PPC6800 / Telus & HTC P4000 / UTStarcom PPC6800 / UTStarcom & Verizon XV6800 Wizard: HTC P4300 / Qtek 9100 / Qtek A9100 / Dopod 838 / i-mate K-JAM / T-Mobile MDA Vario / O2 Xda Mini S / Orange SPV M3000 / Cingular 8125 / Vodafone VPA Compact II Herald: HTC P4350 / Dopod C800 / Dopod C858 / O2 Xda Terra / Vodafone VPA Compact IV Atlas: HTC P4351 / T-Mobile Wing Hermes: HTC P4500 / HTC Z / HTC TyTN / Qtek 9600 / Dopod 838 Pro / Dopod CHT9000 / i-mate JASJAM / O2 Xda Trion / T-Mobile MDA Vario II / Vodafone VPA Compact III / Vodafone v1605 / SFR v1605 / Orange SPV M3100 / AT&T 8525 / Cingular 8525 / Swisscom XPA v1605 / SoftBank X01HT Kaiser: HTC P4550 / HTC TyTN II / Orange & HTC TyTN II / Vodafone VPA Compact V / Vodafone v1615 / SFR v1615 / Swisscom XPA v1615 / T-Mobile MDA Vario III / AT&T Tilt / O2 Xda Stellar / EMobile Emonster S11HT Niki: HTC P5500 / HTC P5520 / HTC Touch Dual / Orange & HTC Touch Dual / SFR & HTC Touch Dual / Vodafone & HTC Touch II / Dopod S600 / T-Mobile MDA Touch Plus / Swisscom XPA Touch Dual / O2 Xda Star Neon: HTC P5530 / HTC P5550 / HTC Touch Dual 850 / NTT DoCoMo FOMA HT1100 Panda: HTC P6300 / Dopod E616 / O2 Xda Argon Sedna: HTC P6500 / Orange & HTC P6500
Niniejszy opis nie stanowi oferty w rozumieniu Kodeksu Cywilnego. Multi-COM Sp. z o.o. zastrzega sobie rwnie prawo do bdw w powy szym opisie produktu. W celu weryfikacji prosimy o kontakt z naszym Dziaem Sprzeda y.
Multi-COM Sp. z o.o. ul. 22-go Lipca 31b 36-100 Kolbuszowa POLSKA tel. 0048172270025, 0048172275045 Infolinia: 0801671717 Skype: Multi-COM

Field Test of Mobile Wi-Fi Terminals in a Wireless City
Petter Stray, Poul E. Heegaard, Thomas Jelle
Department of Telematics Norwegian University of Science and Technology petter.stray@gmail.com, {poul.heegaard, thomas.jelle}@item.ntnu.no

Abstract

We evaluate the achievable voice quality using VoIP on hybrid WiFi/GSM/3G mobile terminals in the citywide wireless network deployed in Trondheim. network loads. Using the network analysis tool IxChariot from Ixia a Performance metrics are gathered from each test and set of Qtek and HTC terminals are tested under dierent conditions and compared up against the already ubiquitous GSM service. This paper is written on the basis of [8].

Introduction

More and more people terminate their land This
Over the last two decades the mobile phone has gone from being a tool for the few to a necessity for the public. line subscriptions, and completely rely on a cell phone for communication.
even though the experienced voice quality on a mobile phone is substantially lower than that on the public switched telephone network (PSTN). People accept this degradation because they value the gain in mobility higher than the loss in quality. Today many people mean that Voice over the Internet Protocol (VoIP) poses a threat to mobile services as we know them. As entire cities become giant wireless zones it is expected that these also support Voice over Wireless Local Area Networks (VoWLAN). If these large scale networks provide the necessary resources to obtain voice quality at a sucient level, people are likely to use this as their primary communication network. The advantages are many; limitless downloading, free Voice over IP using Skype, GoogleTalk a.o. and the possibility to simultaneously send multiple types of trac over the same network. In Trondheim such a citywide wireless network has been deployed on the initiative of the Norwegian University of Science and Technology (NTNU). The network is built with special weight on providing more than enough bandwidth so high capacity services are providable. However, since wireless cities are just starting to be deployed, little research and testing is done on the services and equipment to be used in these networks. Testing dierent mobile Wi-Fi terminals in Wireless
This paper was presented at the NIK-2007 conference; see http://www.nik.no/.
Trondheim under various network conditions can help give an indication of what works best in a system as the one deployed here. Comparing the quality of a mobile voice service up against the quality of the already ubiquitous GSM service can be useful to see if the potential is present to take voice over citywide wireless networks into use.

Evaluating Voice Quality

Comparing the quality of data networks is a complex task, since there are many factors to consider. Dierent applications weight a networks performance metrics in dierent ways. Some applications may require high bandwidth, but are less sensitive to variations in the end-to-end delay. For voice services, a single metric has been established to rate call quality - the Mean Opinion Score (MOS). For Voice over IP, the E-model provides a number comparable to MOS.

Mean Opinion Score

Assessing call quality has traditionally been a subjective task; picking up a telephone and listening to the quality of the voice. The most widespread subjective voice quality metric is the Mean Opinion Score described in the International Telecommunications Union (ITU) recommendation P.800 [7]. The MOS for a call service is calculated from letting a large number of people listen to audio and give their opinion of the call quality on a scale from 1 to 5. description related to it as illustrated in table 1. Each score has a However, if every little tuning As a consequence
adjustment to a phone service requires a large amount of people listening and rating the quality it would be both expensive and time-consuming. E-model. objective measurement techniques have been established, one of them being the
Rating Denition Description
1 Excellent Good Fair Poor Bad a perfect speech signal recorded in a quiet booth intelligent and natural like PSTN telephone quality communication quality,but requires some hearing eort low quality and hard to understand the speech unclear speech, breakdown
Table 1: ITU P.800 MOS conversation opinion scale [7].

The E-model

The E-model was introduced in the ITU recommendation G.107. The E-model uses measured delays and equipment impairment factors to calculate a single scalar, the R factor. The R factor is given by the equation [1]:

R = R0 Is Id Ie (+A)

where:
: unaltered signal, expresses the basic signal-to-noise ratio (SNR)

Is Id Ie A

: impairments that occur simultaneously with the voice signal, such as too
loud speech level : delays introduced from end-to-end : impairment introduced by the equipment : advantage factor, willingness to trade voice quality for convenience
These delay and equipment impairments are inuenced by the data networks one-way delay, jitter and data loss. Implicitly the codec used also inuences the The advantage delay and impairments, especially if a compression codec is used. analysis tool used.
factor is in parenthesis in equation (1) because it is not regarded by the network
Figure 1: Translation from objective R factor to subjective MOS value[4].
The R factor ranges from 0 to 100,
and can easily be translated into a
corresponding MOS value. When a voice conversation is converted to a digital signal and back there is an inherent degradation. This reduces the theoretical maximum R factor with no impairments from 100 down to 93.2 [4]. The translation from R factor values to MOS values is illustrated in gure 1. In the gure you see the R factor values from the E-model to the left, the likely opinion of human listeners in the middle and MOS values to the right.

Performance Criteria

Generally voice quality is considered to be good if it exceeds a MOS value of 3.5. However, according to [3], a normal GSM network only delivers audio with a MOS score between 2.9 and 4.1. The lower bound of this interval is relatively far down in the Nearly All Users Dissatised section of the MOS scale in gure 1. People clearly value the mobility they get from a cellular phone a lot since they accept such a large degradation in voice quality. Telenor, Norway's largest GSM provider, has provided some voice quality They have statistics from their mobile network in six major Norwegian cities. results are summarized in table 2.
measured the share of conversations that obtain a MOS score of over 3.0, their

MOS down-link % over 3.0

MOS up-link % over 3.0 99.40 98.85 100.00 98.18 99.41 98.71
lesund Bergen Stavanger Stor-Oslo Troms Trondheim
98.88 98.36 99.71 97.96 99.41 98.52
Table 2: MOS values from Telenors GSM network [10].
As you can see from the table a very high percentage of the calls in the GSM network qualify for a MOS score over 3, as many as 98.95% of the conversations on average. Considering that most people nd the quality achieved by their GSM phones to be sucient, a lower quality than that in GSM will probably be acceptable for the majority of potential users of a citywide wireless network for voice communication. Since the lower MOS value in GSM is around 3, we assume that a value as low as 2.6 on the MOS score might be acceptable for the basically free voice service that is possible over the packet switched network. We have chosen the lower quality limit 2.6 since this also is the lower border of the Nearly All Users Dissatised group of MOS scores in gure 1.

Test Implementation

The software was installed on a server at Uninett
IxChariot, the analysis tool used in this research, directly gives you the MOS value a conversation achieves.
Trondheim, while small software parts, called
performance endpoints, were installed
on all of the terminals being tested. Two laptops were also used in the tests. One was used to start and stop tests through a remote desktop connection to the IxChariot server, while the other one was used to generate background trac or act as an endpoint in conversations with the mobile terminals. A desktop computer placed at NTNU's largest campus, Glshaugen, was used in the same way. Figure 2 illustrates how the software works. Testing in IxChariot is script-based, so at startup the IxChariot console sends the script to be run to the involved terminals. The endpoints involved can either collect the performance information and transfer it to the console when the test is nished, batch-mode, or they can constantly update the performance to the console, realtime-mode. In this research batch-mode was used while performing stationary tests, and real-time-mode was used in tests involving roaming. This because batch-mode provides the most accurate results since no performance information is transferred in the network during a test. up till that point is reported. If VoIP is tested, the trac between Endpoint 1 and 2 in gure 2 follows standard Real-time-mode is used in roaming because if the connection to an access point is lost during a test no information on the performance
Uninett is developer and operator of the Norwegian research network which provides high capacity Internet connectivity to over 200 educational and research institutions in Norway.
Figure 2: Example of WLAN test setup with Ixia's IxChariot.
VoIP procedure with Session Initiation Protocol (SIP) control plane followed by Real-Time Transport Protocol (RTP) stream and teardown [5].

Test Setup

The tests in this research were performed at dierent locations in Trondheim, and dierent combinations of terminals were used. A total of ve mobile Wi-Fi terminals were tested, one Qtek 9000, one Qtek 8310, two Qtek 8300's, and one HTC TyTN. Voice conversations were set up between the mobiles, between mobiles and a laptop, and between mobiles and a desktop computer. A few tests were also run only involving a laptop and a desktop, in order to see if limitations were located at the terminal or network side. Varying amounts of background trac were also used to see how the voice quality was aected by dierent trac types. Tests were performed using two of the most common voice codecs, ITU-T G.711 and G.729. The two codecs have fairly dierent characteristics. G.711 is the preferred codec in PSTN/ISDN [12], merely digitizing voice and providing a 64 kbps stream. G.729 uses compression as well and provides 8 kbps of output. Since G.729 provides a fairly low bitrate output it is often preferred in wireless voice networks [9].

Results

Most infrastructure producers don't recommend more than six or seven active voice conversations at once on their access points. To test the capacity of the access points in Wireless Trondheim up to ten concurrent conversations were tested at one access point. Figure 3 shows the achieved quality using only a laptop, and how the quality was after switching to mobile Wi-Fi terminals. An increasing amount of consecutive calls was also used to test the network cell call capacity. Testing with from three to ten conversations, you could clearly see that the voice quality dropped below acceptable values when seven calls were active at once. This was the same for both voice codecs. The eect of adding background application data trac to the access point was also tested. Tests were performed using both bandwidth demanding applications

(a) Laptop only.

(b) After introducing mobiles.
Figure 3: 10 G.729 conversations.
as well as ordinary web-browsing trac.
While a high throughput application
generating a little over 3 Mbps of trac caused the voice quality in ongoing conversations to drop with 25 %, simple web-browsing generating a little over 230 kbps of trac caused a drop of 15 % in voice quality.
(a) Without backgound trac.

Figure 4:

(b) With backgound trac.
4 G.729 conversations with background trac.
Figure 4 shows how voice quality in the network deteriorates after adding background trac. The average voice quality in the ongoing conversations drops from 3.34 to 2.82 after introducing background trac in gures 4(a) and 4(b). The yellow-shaded area in both gures indicate the quality that is obtained in GSM networks.

Roaming

Tests on roaming were performed walking with the roaming terminal out of the coverage area of the associated access point and into the coverage area of another AP. Roaming was also performed using both codecs, and the two codecs performed quite dierently. While conversations using G.711 generally held the conversations going longer than those using G.729, the conversations using G.729 maintained higher voice quality.

Terminal Dierences

The tested terminals had fairly dierent performance characteristics. The Qtek 9000 was the most powerful terminal tested with a 520 MHz processor. terminals. This was also reected in the results, where the Qtek 9000 generally outperformed all the other The HTC TyTN was the terminal with the second best prerequisites
to perform well with a 400 MHz processor.
However, the TyTN was unable to
provide good voice quality on the streams going to it. While the streams originating at the TyTN mostly achieved high voice quality, the streams terminating at the terminal would have been completely inaudible. The two Qtek 8300's and the 8310 all performed very similar. On average these three terminals performed better then the TyTn, but worse than the Qtek 9000. Also these terminals had greatly varying quality on streams originating and terminating at the mobile. However, not as diering as the TyTN. Figure 5 shows how the tested terminals vary in performance.
Figure 5: Comparison of terminal performance.
Test in Indoor Enterprise Network
The results obtained in the tests in the Wireless Trondheim network were below acceptable values. The tests performed could not unveil if the cause of poor Trying to gure out what was the cause of poor performance was the terminals or if problem lay in the compatibility between the terminals and the network. performance tests were also performed in an indoor enterprise network at Uninett. The results obtained here were far better. As you can see from gure 4 the voice quality was more uniform at Uninett. Figure 6(a) shows that the voice quality in Wireless Trondheim is constantly jumping between 1 and 4 on the MOS scale, while gure 6(b) shows how the quality obtained at Uninett is only shifting between 3.8 and 4. In both gures the lines averaging between 1.2 and 1.5 are the streams terminating at the HTC TyTN.
(a) Results from Wireless Trondheim.
(b) Results from Uninett.
Figure 6: Two G.711 and two G.729 conversations.
These tests indicate that the poor performance in Wireless Trondheim at least isn't caused by poor processing power at the terminal side.
Isolated Access Point Test
A few nal tests were performed at an isolated access point in Wireless Trondheim. The access point is placed in a park, and is long enough away from other access points for interference to be a problem. In these tests a spectrum analysis was performed as well to monitor trac on the dierent channels in the 2.4 GHz band. As you can see from gure 7 there was only trac on channel 11 in these tests, which is the channel the access point is operating within. the channel 11 area turns completely white. In gure 7(b) you can clearly see that the voice conversation started after approximately 40 seconds where

(a) Radio Frequency (RF) power as a function of frequency.
(b) RF power per 1 second sweep.
Figure 7: 2.4 GHz band activity.
All the terminals performed far better connected to the isolated access point. Although the performance varied when the terminal was far away from the access point at close to it, the overall performance was more similar to the results obtained at Uninett than those obtained in the city. These tests rule out that compatibility issues between the tested terminals and the Wireless Trondheim network are to blame for the performance weaknesses.

Discussion

As the test results show
The call capacity in each network cell is a constraint.
each access point can handle approximately six conversations at once. If application trac is introduced to the access point as well, calls may be dropped and/or the voice quality in the ongoing conversations will drop signicantly. An access point usually has a signaling radius of about 50 meters [6]. This means that when walking down a street in Trondheim you will roam from one access point to another approximately every 100 meters. Walking in one of the main streets in Trondheim on a sunny Saturday afternoon you will most likely see many more than six people every hundred meters talking in their cell phones. Although the density of access points in the city is much higher in such high trac areas, there will probably be many situations where an access points capacity is breached. Reducing the transmit power on the access points can be one solution to the cell capacity problem. By reducing the transmit power each access point covers a smaller
area, which lowers the risk of overloading an AP. In a wireless network covering an entire city this would be a very expensive solution since lowering the range on each of the access points would require the deployment of a signicant amount of new access points. Another solution could be using a seamless roaming system such as the Generic Access Network (GAN). GAN, also known as Unlicensed Mobile Access (UMA), enables a terminal to seamlessly roam between GSM/UMTS and a WLAN if a WLAN comes within reach [11]. If this is combined with a capacity check of the WLAN before connecting to it, this could greatly prevent access point overloads. Quite a few terminals with GAN/UMA support has already reached the market from large cell phone producers such as Nokia and Samsung. For further reading about UMA the reader is referred to [11].

Comparison with GSM

As previously stated, it is unlikely that people will start to use voice applications over the Wireless Trondheim network unless a certain voice quality is maintained. Since almost everybody uses a cellular phone every day, it is natural to evaluate the quality of VoWLAN up against the GSM phone service. As you saw in table 2 as many as 98.95 % of the calls made in Telenors GSM network qualify for a MOS score over 3. Using the G.729 codec on a set of terminals in the Wireless Trondheim network only 77.46 % of the conversations score over 3.on the MOS scale. This means that almost th of the time the voice quality using 4 voice over wireless is below the quality you get using GSM. In the same case only a little over 80 % of the conversations maintain MOS values over 2.6. If however you are unlucky and somebody is downloading a le while connected to the same access point as you it is likely that only 58.16 % of the conversations maintain a voice quality over 3.0. At the same time, only 62 % of the time the conversations maintain voice quality above the minimum MOS level of 2.6 set in section 2. As the tests from Uninett show the terminals behave dierently in an isolated indoor environment where interference is deliberately counteracted upon. The tests performed connected to an isolated access point in Wireless Trondheim support this as well. It seems obvious that the mobile terminals' radios are far to poor to enable time critical applications to function well in a noisy environment, with interference caused by private access points, cordless phones and microwaves [2]. Data applications may well function because retransmission of lost packets is acceptable, but for the time critical application retransmissions are seldom an option. The situation will presumably not improve until mobile hand held devices' radios improve. This can be by improving the current 802.11b or g radios sitting in them already, or that the terminals get support for IEEE 802.11a, while, at the same time, private actors do not install private 802.11a networks. If all 802.11a access points are in the control of Wireless Trondheim a mostly noise free environment can exist in the 5 GHz band, hopefully improving the performance of mobile terminals in a voice over WLAN session.

Conclusion

Using the network analysis
The objective of this research was to perform a series of eld experiments on Voice over IP in the Wireless Trondheim network.
tool IxChariot, Voice over IP was tested on a set of Wi-Fi enabled Qtek and HTC terminals under varying network conditions. Through tests of the available terminals, under dierent network compositions and loads, the overall quality of the service was assessed. The quality of the service was also compared up against results from GSM in order to see if a voice service over the Wireless Trondheim network can compete with GSM. One of the main ndings through the experiments was that in a city like Trondheim there is far too much interference for a mobile terminal to obtain a stable voice conversation with good quality. Todays mobile Wi-Fi terminals operate in the 2.4 GHz band, alongside private wireless networks, cordless phones and microwave ovens. A densely populated city center therfore contains a lot of interfering signals. On average, each of the over 100 access points in the Wireless Trondheim network sees 12 interfering access points not belonging the network. This naturally inuences the clarity of the received signal at the terminals as well. The obtained voice quality has been measured on a ve point scale, called the mean opinion score (MOS), where 1 is poor and 5 is excellent quality. Nearly all calls made in the GSM network score over 3.0 on the MOS scale. Taking this, among other factors, into consideration the criteria for acceptable quality for the voice over Wireless Trondheim service was set to 2.6 on the MOS scale. The tested terminals have performed dierently, however, the obtained quality in the tests has been constantly lower than the criteria set for the service. The voice quality in the network degrades further when adding data trac to the network. The presence of ordinary web browsing trac causes the voice quality of ongoing conversations to drop almost 0.5 on the MOS scale. An indirect factor also aecting the experience of voice over the Wireless Trondheim network was the battery life of the mobile terminals. Connected to a wireless network the terminals' batteries only last a few hours. Considering that most people have wireless networks available both at work and at home, it is likely that a practically free voice over wireless service will be desired to use as ones primary communication medium. If this involves recharging your terminal several times a day it is not likely to catch on in the near future. Without any prioritizing of latency sensitive trac the shared wireless network seems unsuitable for VoIP services. After testing several dierent terminals in the Wireless Trondheim network the conclusion is quite clear, voice over the citywide wireless network is not yet ready for the masses.

References

[1] ITU-T Studygroup COM12. The E-model. (online), 2007. URL: http: //www.itu.int/ITU-T/studygroups/com12/emodelv1/introduction.htm.
[2] Jim Geier. The state of Wireless LANs. Sponsored Exclusively by intel.
NetworkWorld's Special Report, 2004.

http://www.inline.

White Paper,
[3] inline Systems AB. Voice Quality. (online), 2005. URL:
se/products_vq_voice_quality.html.
[4] Ixia. 2005. Assessing URL: VoIP Call Quality Using the

E-model.

http://www.ixiacom.com/library/white_papers/display? skey=voip_quality.
[5] Ixia. IxChariot. (online), 2007. URL:
http://www.ixiacom.com/products/display?skey=ixchariot.
[6] Thomas Jelle. Wireless Trondheim - 1:2006. Norwegian University of Science and Technology (NTNU), Jan 2006. [7] Telecommunication Standardization Sector of ITU. Methods for subjective
determination of transmission quality. ITU Recommendation, 1998. [8] Petter Stray. Field Test of Mobile Wi-Fi Terminals in a Wireless City. Master's thesis, Norwegian University of Science and Technology (NTNU), 2007. [9] Ready Technology. Open Source G.729 and G.723.1. (online), 2007. URL:
http://www.readytechnology.co.uk/open/ipp-codecs-g729-g723.1/.
[10] Telenor. Tale statistikk telenor area mai (Norwegian). Private communication, 2007. [11] UMA Today. UMA Technology. (online), 2007. URL:
[12] Trond Ulseth and Finn Stafsnes.

http://www.umatoday.

Telektronikk 1.2006, 2006.
VoIP speech quality - Better than PSTN?