A software tool (based on Matlab) has been designed and implemented for the topology and capacity design of a heterogeneous wireless on-board network, composed by UMTS, WLAN IEEE 802.11b and BluetoothTM access technologies. Basically, this tool has the following capabilities:
· To generate traffic flows in uplink and downlink applying traffic models for different applications (voice, www, ftp, etc) based on traffic input parameters provided by the user and a certain selected aircraft type and flight duration. This gives the flexibility to simulate different traffic scenarios that can be based on realistic assumptions (e.g. a market survey). Furthermore, in order to simulate a real system, the traffic flows are organised in sessions and each session is randomly mapped to a passenger. This feature is relevant in order to locate active transmitters and receivers on board in every instant of time during the flight. Figure 1 shows an example of the traffic flows generated with this tool for WLAN access, in a scenario where web, email and streaming traffic over WLAN are supported in a long range aircraft; the start and stop instant of each session can be observed and the corresponding passenger and application can be identified for each session. It should be noted that in this figure an average throughput is assumed during the whole session duration, though the topology and capacity planning tool is able to store instantaneous bit rate as well.
Figure 1: Traffic activity per passenger and application during the flight in a scenario supporting www, email and streaming over WLAN (assuming average bit rate).
· To define a default network topology that is able to support the generated traffic flows, taking into account the areas of the cabin where antennas are allowed to be installed. The software simulates the mobile terminals and access points on board and their relative distances. Based on propagation models derived from the progress of WP4200, the software estimates the received power strength at each cabin coordinate of a defined height plane of interest from all active transmitters on board. The transmission and reception characteristics of mobile terminals and access points are configurable by the user in an access technology basis. As an example, Figure 3 shows the results of such propagation calculations for access points and mobile terminals of the UMTS and WLAN simulated on-board network in a short range aircraft: A319. It should be noted that the green seats in the detailed cabin layout (left side of the figure) are correlated with the received signal strength from the UMTS mobile terminals (part (c) of the figure). The red rectangle above the detailed layout corresponds to the area represented by the simplified layouts in parts (a), (b), (c) and (d) of the figure.
· To identify potential interferers to each active receiver on board, differentiating inter- and intra-system interferers. With this, the cumulative instantaneous interference to each receiver is estimated and instantaneous CIR values are derived.
· The software provides tools for the estimation of performance figures, such as BER in the special case of interference between IEEE 802.11b and BluetoothTM. Based on the comparison between the estimated performance figures and defined performance thresholds, the user is able to set a feed-back loop that optimizes the topology and capacity design in order to enhance those parameters that show poor performance, affecting the least possible the performance of satisfactory parameters. Examples of possible countermeasures are the addition of access points, the reconfiguration of the topology, the substitution of point antennas by leaky lines or the increase/decrease of transmit power of on-board transmitters. Criteria for the settings of each loop are provided in [D13], discussing the implications of possible countermeasures to different performance parameters.
Figure 2: Representation of UMTS active users in the cabin layout and received signal strength level at the ear height in when the transmitter is: a) a UMTS access point, b) a WLAN access point, c) UMTS mobile terminals and d) WLAN mobile terminals.
Three scenarios have been defined in [D13] in order to provide the topology and capacity design of three exemplary situations of interest:
· The first scenario represents a short range flight of 90 minutes in an A319, with relatively high voice traffic load over UMTS access and low email traffic load over WLAN. This scenario aims at emulating a short flight where the use of the mobile phone for short voice calls is more expected than data traffic over WLAN.
· The second scenario aims at comparing the WLAN network performance of scenario 1 with the case that BluetoothTM interferers are present on board, affecting the WLAN traffic by increasing the WLAN BER. In this scenario, the UMTS network performance is not of further relevance; therefore, the corresponding traffic load has been relaxed, in order to reduce the simulation time.
· A third scenario represents a long range flight of 270 minutes (4.5 hours) in an A330-200 with lower voice traffic load over UMTS but higher data traffic over WLAN, mixing several applications: www, email and streaming, since in a longer flight it is more expected that the passengers take the time to use their laptops. Furthermore, the presence of BluetoothTM interferers in this scenario is as well considered. In this scenario, a high traffic load (especially in downlink) has been simulated in order to stress the satellite capacity in comparison with the on-board networks’ capacities.
The most relevant results of this topology and capacity design are summarized in the following. Provided that the coverage conditions are very different in small and large aircrafts, the solutions to achieve the required coverage in each kind of aircraft are also different: for a small aircraft, the required coverage is achieved with one access point per access technology using point antennas, whereas for a large aircraft the use of point antennas is very inefficient. In particular, it has been observed in [D13] that even with four access points the cabin shows coverage gaps for WLAN. A better solution for the large aircraft is the use of a leaky line, in which case one access point is sufficient for coverage purposes. Figure 3 shows with a colour code the distribution of received signal strength from a WLAN access point across the cabin at ear height when using a point antenna (part a) and leaky line (part b) in a simplified aircraft layout; on the left side of figure, the detailed layout can be observed, where the red and green squares represent inactive and active passengers, respectively; the area marked with a red rectangle corresponds to the area plotted in the simplified layout. Observing the simplified layouts, it is obvious that the signal strength is more efficiently distributed using leaky lines provided the sever propagation conditions in a large aircraft. Furthermore, other advantages of applying the leaky line solution are the avoidance of inter-cell interference and the lower cost, since only one antenna is required.
Figure 3: Comparison of received signal strength distribution from a WLAN access point across the aircraft cabin (in simplified cabin layout) when using: a) point antenna and b) leaky line.
Given the traffic load generated in the different scenarios, the in-cabin network can provide sufficient capacity for WLAN and UMTS networks with one access point per access technology. In what the satellite link is concerned, the defined traffic load for the short range aircraft can be served by a medium rate satellite system, such as GAN. That is not the case of the long range aircraft scenario. By including web and streaming traffic, the satellite downlink is overloaded very often with such medium rate satellite system. Even using a broadband system, such as B-GAN, the blocking in downlink exceeds the allowed threshold. This result suggests the possibility to deploy an on-board server that relaxes the satellite capacity demand, in particular for downlink only traffic. The deployment of streaming through satellite would be only realistic with such capacity availability if the content is broadcasted in the cabin, but not for individual downloads, which would overload easily the downlink due to the relatively high data rate demands for streaming traffic.
The coexistence between WLAN and BluetoothTM devices on board is limited to specific scenarios. It has been proved that in the case that the BluetoothTM networks on board are PANs, the coexistence in a short range aircraft degrades the BER of both networks within acceptable margins. In particular, the BluetoothTM connections experiment low BER degradation due to the very short duration of the simulated WLAN sessions; on the other hand the WLAN sessions receive a desired signal level good enough to be protected against collisions. In the long range aircraft, the severe propagation conditions allows for a good BER performance in the BluetoothTM connections only in very short distances, whereas the WLAN sessions have a good interference protection due to the good signal quality provided by the uniform distribution of the signal strength through the leaky line. It should be noted that these results in coexistence are not applicable to all possible network topologies. In particular, it is expected that the deployment of access points for BluetoothTM piconets will be especially critical.