The NDT allows broadband subscribers to accurately measure the qualitative features of their connections and presents them in an impartial manner. It should, however, be noted that:
To ensure a reliable measurement it is essential that (a) no other application is “running” on the user’s computer or local network, which would generate network traffic and (b) there should be no form of temporary network problem on the path to the server.
It should be noted that if other computers are also generating traffic on the user’s local network, this may negatively affect the accuracy of a measurement. The user in any case is advised to ensure that the network connection is as inactive as possible before proceeding to measure.
Using WiFi may affect the accuracy of the measurement. For the most widely used protocol, 802.11g, depending on the quality of the signal, the available WiFi bandwidth varies from 6 to 54 Mbps, and the wireless part of the connection operates in half-duplex mode. Therefore, if there are other devices connecting to the WiFi router (e.g. a smartphone), or the signal is weak, the WiFi speed could be lower than the speed of the wired interface, and as a result the overall speed would be limited by the WiFi speed. This rarely occurs in the case of a wired connection, where the bandwidth is always 100Mbps full duplex (if not more). It is therefore advisable to measure from a computer connected by cable to the domestic router.
Firewalls, heavy use of the computer’s central server and even faults on the computer network cable, may affect measurement quality.
No. The measurement uses TCP/IP as underlying protocols, but does not include this overhead in the throughput calculation. This results in lower speed, but is closer to the throughput an application would see.
It isn’t the same in all cases. It varies according to the WAN protocol used (e.g. PPP or HDLC), as well as application overhead, overhead for encryption, etc.
The user will always get less than the nominal speed he/she purchased as the nominal speed is only a theoretical limit at L2 (Data link layer). The actual speed of connection to the Internet depends on many factors, such as the distance from the local exchange or cabinet, noise levels, the degree of congestion in the provider’s network, as well as the internal network at the subscriber’s premises (wired/wireless connection to the router, existence of other devices in the same local network, cross-traffic from other applications).
Interested users can visit https://www.increasebroadbandspeed.co.uk/2012/graph-ADSL-speed-versus-distance for more details on the variation of the downlink connection speed depending on the distance from the exchange, for an ADSL line.
Nevertheless, it is very difficult to properly account for all factors that affect Internet speed and predict the expected speed without conducting any measurements.
The NDT employs two inter-related sets of protocols: NDTP-Control and NDTP-Tests. Both of them are layered over TCP.
By analyzing the measurement packets (data and acknowledgement) sent during the client user computer-server communication. The user sees the average values based on all the exchanged packets. The difference between the two extreme delay values constitutes the delay variation (jitter), whilst packets that were lost and had to be retransmitted are included in the percentage loss.