Know Your Neighbour: Packet Loss Correlation in IEEE 802.11b/g Multicast

Abstract

In this paper we present a number of wireless measurements conducted on mobile phones using WLAN IEEE802.11b/g in a multicast scenario. When designing efficient error recovery mechanisms for reliable multicast protocols, it is necessary to have good a understanding how channel errors occur and preferably to have a network model that describes these errors well. The measurement results in this work are not sufficient to create an accurate error model, nevertheless they provide two important results. First, the assumption of independent package loss, commonly assumed in simulations and analytical work, does not hold. In all measurements conducted the packet loss, observed at different devices, indicated some level of correlation. Therefore error models used for simulations and analytical evaluation should take this finding into consideration. Secondly, the packet loss probability observed at mobile devices of the same brand and model, exhibited large differences. A uniform packet error probability will therefore not be a satisfactory error model, even when considering homogeneous networks.

Impact of device unavailability on the reliability of multicast transport in IEEE 802.11 networks

“…We believe that this is the case in [9], where the device with the best link quality experiences the highest loss rate. The same observation is made in [10] where the authors confirm that identical devices perform quite differently in terms of packet error rates. This conclusion is reached using a multicast group of 9 collocated mobile phones and an IEEE 802.11b network.…”

“…There has been intensive work on analyzing the different causes of packet loss in wireless networks [6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24]. The underlying motivation is to understand the reliability of the wireless link and the variation of available bandwidth.…”

Synchronized multimedia streaming on the iPhone platform with network coding

“…Here we assume that the first-tier neighbors have a 10 percent loss rate, while the second-tier device has a packet error rate of 50 percent. Those values are reasonable and were reported in [8]. If communication is only allowed on the blue links (i.e., the originating device is transmitting packets), the overall time until all devices receive the file depends primarily on the second-tier neighbor.…”

Performance evaluation of coded meshed networks

“…More importantly, the interference confirms that the loss rate is highly dynamic over time, despite the static nature of the setup. Table I the set of receiving nodes, indicating that an assumption of independent channels might not apply, a notion supported by the efforts in [15] Due to the more efficient channel modulation and coding schemes, the case of unicast transmissions without the MAClayer retransmission in Fig. 3 seem very unreliable with up to 55% packet loss.…”