Directional Antennas
3.6 Simulation Results
Based on the proposed code/time slot assignment scheme, we simulate the signal-to-interference ratio (S/I) and call blocking performance in a TDD-CDMA system with
Figure 3.13: The scheme of time slot allocation.
19 cells as shown in Fig. 3.11. To demonstrate the advantage of using directional antennas in TDD-CDMA system, four kinds of slot assignment schemes are compared:
• Scheme I: the global control on the switching point setting in the TDD-CDMA system with omni-directional antennas;
• Scheme II: the distributed control on the switching point setting in the TDD-CDMA system with omni-directional antennas, where every cell has different switching point setting;
• Scheme III: the proposed virtual cell-based slot allocation algorithm in the trisector TDD-CDMA cellular system;
• Scheme IV: the sector-by-sector based switching point setting in the trisector TDD-CDMA cellular system.
In our simulation, we evaluate both the bit energy to the interference den-sity ratio Eb/I0 and call blocking performances in the TDD-CDMA system with directional antennas and those with omni-directional antennas, where Eb/I0(dB) = S/I(dB) + P G(dB). Here we consider a processing gain P G = 10 dB. Assume that
the up/downlink bandwidth requirements in each sector are different as shown in Table 3.6. Let each cell have the same number of codes and time slots, i.e., a frame containing 8 time slots and 9 sets of orthogonal codes. Furthermore, we assume that every virtual cell is allowed to manage its own radio resource, To eliminate the boundary effect in simulation, we also adopt the wrap-around technique in a 19-cell environment.
Figure 3.14 compares the uplink Eb/I0performance of the four aforementioned slot assignment schemes. In the figure, there are two groups of curves. The curves in the right part are the Eb/I0 performance for a system with directional antennas, and the left part is that for a system with omni-directional antennas. It is obvious that both Schemes III and IV with directional antennas outperform Schemes I and II with omni-directional antennas. Focusing on the curves of Schemes III and IV, we find that the Eb/I0 performance of Scheme III is slightly better than Scheme IV. This is because the intercell interference is smaller in Scheme III. Though in Scheme IV the intercell interference can also be restricted in a small area, if the switching point settings of the three neighboring sectors in a virtual cell are different, the cross-slot interference among the three sectors can still degrade the 90th percentile of Eb/I0 performance by 1.5 dB compared to Scheme III. Because of the same reason, the Eb/I0 performance of Scheme I (i.e., the omni-directional antenna case with the global control on the switching point setting) is better than Scheme II (i.e., the omni-directional antenna case with local setting). Thus, in the system with omni-directional antennas, it may be necessary to set the same switching point in all cells to avoid cross-slot interference.
Figure 3.15 compares the downlink Eb/I0 performances of the aforementioned four time slot allocation schemes. In the figure, one can see that the Eb/I0 perfor-mances of Schemes III and IV (the curves in the right part) outperform Schemes I and II. For example, the 90th percentiles of Eb/I0 for Schemes III and IV are 5 and 1.5 dB, respectively, while the 90th percentiles of Eb/I0 for Schemes I and II are -9
Table 3.2: Uplink and downlink traffic requirement in each sector of Fig. 3.11.
Sector 1 Sector 2 Sector 3
Required Required Required Required Required Required Cell UL Slot DL Slot UL Slot DL Slot UL Slot DL Slot
A 2 5 2 6 2 4
B 2 5 2 3 2 6
C 5 2 2 5 3 3
D 2 4 2 2 3 4
E 5 3 2 4 3 2
F 2 5 3 5 2 5
G 2 4 2 6 2 5
H 4 2 2 5 2 3
I 2 4 2 4 2 5
J 3 3 2 3 2 2
K 3 4 4 3 4 2
L 5 2 2 5 2 4
M 3 5 2 4 2 5
N 2 5 2 4 2 5
O 2 3 2 4 2 4
P 2 5 3 3 4 3
Q 5 2 3 4 4 3
R 4 2 5 2 2 5
S 2 4 3 5 2 4
and -10 dB. The performance improvements of Schemes III and IV over Schemes I and II are explained as follows. The major cross-slot interfering signals are from all the surrounding six cells in Schemes I and II, while there are only three sectors within a virtual cell causing cross-slot interference in Schemes III and IV. Besides, one can see that the Eb/I0 performance of the two omni-directional cases are very close. This is because in Schemes I and II most interfering signals during downlink transmissions are from base stations and their interference scenarios are similar.
Figure 3.16 shows the call blocking performance of the four different alloca-tion schemes. We can see that the proposed Scheme III has the same call blocking performance as the omni-directional antenna case, i.e. Scheme II. Recall that Scheme III aggregates all the codes/time slots of the three sectors in a virtual cell and then assigns these time slot/codes to the covered mobile terminals, while Scheme II as-signs codes/time slots to mobile terminals in a cell with omni-directional antenna.
Since a virtual cell has the same coverage area and the same number of codes/time slots as an omni-cell, the call blocking probability of the two schemes perform closely.
However, Scheme II has poor radio link performance as shown in Figs. 3.14 and 3.15. In Scheme I, an optimal switching point setting is required for the global up-link/downlink bandwidth ratio among all the cells within the entire system. If adopt-ing a global uplink/downlink bandwidth ratio as in Scheme I, each cell may sacrifice its actual traffic requirement, thereby causing higher call blocking than Schemes II and III. The blocking rate of Scheme IV is very high due to the decrease of trunk-ing efficiency. Observtrunk-ing from Figs. 3.14, 3.15, and 3.16, we can conclude that the proposed virtual cell-based resource allocation in TDD-CDMA systems with trisector cellular architecture can have the best Eb/I0 performance, while maintaining good blocking rate performance.
Figure 3.14: Uplink Eb/I0 performance of slot allocation schemes in both omni-directional and trisector cellular system, where Scheme I is the global setting in omni case, Scheme II is the local setting in omni case, Scheme III is the proposed virtual cell-based case, and Scheme IV is the sector based setting in directional case.
Figure 3.15: Downlink Eb/I0performance of slot allocation schemes in both omni-directional and trisector cellular system, where Scheme I is the global setting in omni case, Scheme II is the local setting in omni case, Scheme III is the proposed virtual cell-based case, and Scheme IV is the sector based setting in directional case.
Figure 3.16: The blocking rate comparison of TDD-CDMA system between four different setting, where Scheme I is the global setting in omni case, Scheme II is the local setting in omni case, Scheme III is the proposed virtual cell-based case, and Scheme IV is the sector based setting in directional case.
52