System Parameters Adjustment

To observe the throughput gain from the deployment of HeNBs in the cell-edge region, we observe throughput of MeNB in the pure macro network and that of HeNBs in the macro-femto networks. In the pure macro network, there is no other deployed eNBs, and then MeNB could use the total bandwidth without interfered by other eNBs. Moreover, HUEs and MUEs are sub-scribed to MeNB, while HUEs are still deployed inside the apartment, and MUEs are deployed outside the apartment. In this scenario, MeNB only performs the PBD and PRA algorithms, and the biased priority value for HUE,u,is set to be 0.

Figure 4.1 shows the throughput of UEs, throughput of HUEs, and throughput of MUEs in the macro-femto and pure macro networks. It can be observed that the throughput of UEs, throughput of HUEs, and throughput of MUEs in the macro-femto networks are larger by more than 47.5%, 54.4%, and 43.9%, compared to those in the pure macro network, as the total traffic intensity is larger than 0.58. Besides, HUEs have throughput degradation of 13.2%, and MUEs have throughput increase of 24.2%, when the total traffic intensity is larger than 1.17. The first phenomenon is due to the fact that the cell-edge UEs have bad channel condition, and MeNB cannot transmit data to them with high modulation order. The second phenomenon is because HUEs usually have worse channel condition than MUEs, and there is insufficient resource to serve all UEs as the total traffic intensity increases. Then, MeNB would tend to serve MUEs prior to HUEs.

36

Figure 4.1 (a) Throughput of UEs, (b) throughput of HUEs, and (c) throughput of MUEs in the macro-femto and pure macro networks.

Since the number of serving UEs of hybrid access HeNB is limited by the Kmax, and thus MUEs could establish connection to hybrid access HeNB, when the number of serving UEs of hybrid access HeNB is smaller than the Kmax. Moreover, different values of Kmax would affect the performance of the CPRM scheme, and thus we would observe the performance of the CPRM scheme in Kmax of 10, 15, and 20.

Figure 4.2 illustrates the throughput of UEs, throughput of HUEs, and through- put of MUEs. In the total traffic intensity of 1.46, we can find that the throughput of UEs, throughput of HUEs, and throughput of MUEs in Kmax of 15 are slightly smaller by 0.9%, 1.4%, and 0.7%, compared to those in Kmax of 10. Besides, throughput of HUEs is similar in Kmax of 15 and 20, while throughput of UEs and throughput of MUEs in Kmax of 15 are slightly larger by 0.5% and 0.8%, compared to those in Kmax

of 20. This consequence is because the CPRM scheme tends to serve the urgent UEs

37

first to satisfy the QoS requirements, and then serve the UEs with better channel condition to maximize the throughput of UEs, while the CPRM scheme in smaller Kmax would have less urgent UEs.

Figure 4.2 (a) Throughput of UEs, (b) throughput of HUEs, and (c) throughput of MUEs.

Figure 4.3 presents the packet dropping rate of voice and video UEs, while Figure 4.4 presents the packet dropping rate of voice HUEs and MUEs, as well as that of video HUEs and MUEs. It can be seen that the packet dropping rate of voice and video UEs in Kmax of 10 is smaller than that in Kmax of 15, while the packet dropping rate of voice and video UEs in Kmax of 20 violates the packet dropping rate requirement in the total traffic intensity of 1.46. Besides, the packet dropping rate of video UEs in Kmax of 10 increases rapidly as the total traffic intensity increases. The first phenomenon is due to the fact that there would be less urgent UEs in smaller

38

Kmax, and the CPRM scheme in smaller Kmax would have more resource to guarantee the delay tolerance requirements of voice and video UEs. However, the CPRM scheme in Kmax of 20 accepts too many voice and video MUEs, and has insufficient resource to guarantee their delay tolerance requirements. However, the second phenomenon is because the video packet size is large, and the video MUEs might be served by HeNB that cannot transmit data with high modulation order. Furthermore, those HeNB might be unable to provide the basic service to video MUEs as the total traffic intensity increases

Figure 4.3 Packet dropping rate of (a) voice and (b) video UEs.

39

Figure 4.4 Packet dropping rate of (a) voice HUEs and MUEs, as well as (b) video HUEs and MUEs.

However, since the throughput of UEs in Kmax of 15 is similar to that Kmax of 10, and the CPRM scheme in Kmax of 15 can satisfy the QoS requirements of all traffic types, as well as it does not have rapid increasing packet dropping rate of video MUEs, we set Kmax to be 15 and evaluate the performance. Besides, we would adjust the deployment ratio of femto block, rd, and then evaluate the performance.

Figure 4.5 exhibits the throughput of UEs, throughput of HUEs, and throughput of MUEs. It can be learned that the CPRM scheme in rd of 0.2 has the smallest throughput of UEs, throughput of HUEs, and throughput of MUEs. Besides, it has dramatic decreasing throughput of MUEs as the total traffic intensity exceeds 0.88.

However, the CPRM scheme in rd of 0.6 has larger throughput of UEs, throughput of HUEs, and throughput of MUEs, compared to those of the CPRM scheme in rd of 0.4, after the total traffic intensity is 0.88. The first consequence is because there are less

40

HeNBs in smaller rd, and thus less UEs would be served. Moreover, the second consequence is also because there are too many UEs for HeNBs to serve in rd of 0.2, and then MUEs have dramatic decreasing throughput. On the other hand, the third consequence is due to the fact that there are more HeNBs in rd of 0.6, and thus more interference would be generated. Therefore, the CPRM scheme in rd of 0.4 can achieve larger throughput to UEs than the CPRM scheme in rd of 0.6, before it can only achieve similar throughput of HUEs to the CPRM scheme in rd of 0.6, but smaller throughput of MUEs than the CPRM scheme in rd of 0.6.

Figure 4.5 (a) Throughput of UEs, (b) throughput of HUEs, and (c) throughput of MUEs.

Figure 4.6 depicts the packet dropping rate of voice and video UEs, while Figure 4.7 depicts the packet dropping rate of voice HUEs and MUEs, as well as that of video HUEs and MUEs. We can notice that the CPRM scheme in rd of 0.2 violates the

41

packet dropping rate of voice and video UEs and that of MUEs in the total traffic intensity of 0.88. Besides, the CPRM scheme in rd of 0.6 has smaller packet dropping rate of voice and video UEs, compared to that of the CPRM scheme in rd of 0.4. The first phenomenon is because there are not enough HeNBs to serve the MUEs in smaller rd, and then voice and video MUEs have dramatic increasing packet dropping rate as the total traffic intensity increases. Moreover, the second phenomenon is due to that there are more HeNBs in larger rd, and thus more MUEs could be served.

Figure 4.6 Packet dropping rate of (a) voice and (b) video UEs.

42

Figure 4.7 Packet dropping rate of (a) voice HUEs and MUEs, as well as (b) video HUEs and MUEs.

Figure 4.8 shows the average transmission rate of HTTP UEs, as well as that of HTTP HUEs and MUEs. We can discover that the CPRM scheme in rd of 0.2 would violate the minimum transmission rate of HTTP UEs as the total traffic intensity is larger than 0.88, and violates that of HTTP HUEs and MUEs in the total traffic intensity of 0.88 and 1.17, respectively. Furthermore, the CPRM scheme in rd of 0.6 can achieve the largest transmission rate of HTTP UEs. These consequences can be explained as in Figure 4.6 and 4.7.

43

Figure 4.8 (a) Average transmission rate of HTTP UEs, as well as (b) HTTP HUEs and MUEs.