Frequency Planning Techniques for Multi-hop Cellular Systems

The first technique proposed in existing literature [44,45] is shown in Figure 20. Its basic idea is to allocate the frequency channels to each MR-BS first, where the channels can only be reused in the MR-BSs in different clusters [15]. Then the RSs in each MR-cell reuse part of the frequency channels allocated to the MR-BSs in adjacent cells. In Figure 20, this technique will divide the total frequency channels into K blocks, where K is the frequency reuse factor defined in Section 2.1 and

4

K = is considered in this example. Then the MR-BS in each cluster will be

RS will select one of the frequency blocks which correspond to longest co-channel distance with the MR-BSs allocated by the same block. Then the RS will reuse 1 6 of the frequency channels in that block, so as to prevent interference with the nearby RSs which select the same block.

Figure 20. The frequency planning technique proposed in literatures [44,45]

The second frequency planning technique [46-49] is shown in Figure 21. Its basic idea is applying the concept of reuse partitioning to divide the MR-cell into inner cell and outer cell then perform frequency planning separately, where the inner cell is served by MR-BS and the outer cell is served by the RSs. In the first step, this technique will partition each MR-cell into inner cell and outer cell. In the second step, it will divide the access zone into two resource-zones to serve the user located in inner cell and outer cell respectively, like the example given in Section 2.1.1 with Z = . 2 In the example shown in Figure 21, the reuse factor K = is considered for the inner 1 cell and K = is considered for the outer cell. In the third step, the corresponding 4 resource-region will be allocated to MR-BS and RSs according to the frequency reuse factor considered in each sub-zone. For the RSs in the same outer cell, the frequency channels will be exclusively allocated to each of them.

In the first technique, the reuse factor considered for MR-BS frequency planning aims to guarantee the MSs located at MR-cell boundary can be served by MR-BS.

However, the MSs located at cell boundary are usually served by RS in multi-hop cellular network, and the MSs directly served by MR-BS are usually located around cell center. For the MSs around cell center, the interference from adjacent cells is much lower the one received by the MS located at MR-cell boundary. Therefore, the frequency reuse factor considered by the coverage served by MR-BS can be lower for more available radio resource, i.e. this technique may result in certain overdesign and cannot fully utilize the scare resource in the system.

The second technique mitigates the aforementioned problem by applying the concept of reuse partitioning (see Section 2.1.1) to partition each MR-cell into inner and outer cells. The inner cell is directly served by MR-BS and more aggressive frequency reuse (e.g. K = in Figure 21) can be used for more resource in access 1 links. In the outer cell served by RSs, conservative frequency reuse (e.g. K = in 4 Figure 21) can be applied to prevent the significant interference from adjacent MR-cells. However, this technique does not allow frequency reuse among the RSs in the same MR-cell and there are still rooms to improve this technique.

In order to utilize the radio resource more efficiently, a new and simple frequency planning technique based on the “sub-cell” concept is proposed in this section. This technique aims to perform more aggressive frequency reuse in multi-hop cellular systems without the damage on coverage guaranteed.

The idea of the proposed “sub-cell” based frequency planning concept and the simple planning procedure are depicted in Figure 22. By dividing each MR-cell into multiple (i.e. 7 in this example) sub-cells served by MR-BS or RSs, the first step is to reorganize the cellular network and define the cluster from sub-cell’s point of view.

the cluster.

5.1.2 Simulation Results A. Simulation Environments

A downlink two-hop cellular network is simulated in Manhattan-like environment, where 6 RSs are deployed within the coverage of an MR-BS. Total of 19 cells with three sectors per cell are considered, and the location of each RS locations is as depicted in Figure 19. The MR-BS and RSs are deployed above rooftop, so that the relay links have line-of-sight (LOS) condition. The access links between MR-BS and MS are assumed to be non line-of-sight (NLOS). For the access links between RS and MS, LOS is assumed if they locate on the same street and the distance is less than 150m. The path-loss and shadow fading models are referenced from the multi-hop relay system evaluation methodology in [41,42]. The total transmit power of each RS is 10 Watts, which is derived by the link budget in Table 5. The total transmit power of each MR-BS is 4 Watts when simulating the second technique and the proposed technique, which is derived by modifying the antenna gain and fading margin of the link budget in Table 5. For the first frequency planning technique, the transmit power of each MR-BS is 16.279 Watts to cover the entire MR-cell, which is derived by the link budget in Table 4.

Two frequency reuse methods for the MR network are simulated to investigate the benefit by proposed RS deployment method. The first method does not reuse the frequencies within the same MR-cell, where the orthogonal frequency channels are allocated to each relay link/access link in relay zone/access zone respectively. In this case, there is no intra-cell interference. The second method is to reuse the frequency channels within the same MR-cell, the channels for relay links are reused in each sector and the ones for access links are reused by every RSs and MR-BSs.

At the beginning of simulation, MSs are generated by Poisson process and randomly located on the street. During the simulation, the MS moves along the streets and communicates with an RS or the MR-BS based on the received signal quality. The modulation and coding scheme is adjusted on a frame-by-frame basis according to signal quality. For each frame, the number of bits (B_A) successfully received in access links is recorded, which will be restricted by the available radio resource determined by different frequency planning technique and the access zone ratio η. The ratio η is defined as the ratio of the radio resource allotted to downlink access zone over the total radio resource available in downlink sub-frame. In addition, the number of bits able to be relayed in each frame (B_R) will be calculated based on the ratio η. Therefore, the number of bits successfully received in each frame will be

( )

min B B_A, _R . At the end of simulation, the cell throughput is calculated by dividing the number of successfully received bits in access links by the overall simulation time.

B. Simulation Results

Figure 23 show the system capacity by given different access zone ratio and under different frequency planning technique. It shows that the proposed technique can achieve to highest capacity in each case. Compare with the first technique, the major improvement by the proposed technique is prevent the overdesign on the frequency reuse factor for the MR-BS sub-cell and use lower K to have more radio resource for access links. In addition, the proposed technique can explore more intra-cell frequency reuse opportunity with respective to the first and the second techniques.

Figure 23 also shows that a tradeoff will exist by giving different η. Larger η can provide more radio resource for access link to serve MSs, but it will also result in less radio resource to forward the user traffic. The optimal tradeoff will exist when the

aggregate traffic in relay links is equal to the traffic in access line. In figure 23, the 62.5%

η = is closest to the optimal point for the second and the proposed technique, and η =75% is the best choice for the first technique.

Figure 23. System capacity by given different access zone ratio (η)

Selecting the highest capacity for each technique in Figure 23, Figure 24 shows that improvement can be achieved by the proposed technique. Compare to the first one, the proposed technique performs 137.75% capacity improvement [51]. Compare with the second one, the proposed technique performs 60.87% capacity improvement. Note that such significant improvement on system capacity is achieved without losing the guarantee on coverage. The coverage guarantee defined here is

(

)

P SINR< dB <

, and Figure 25 shows that all the techniques are compared under the guaranteed

coverage.

Figure 24. System capacity under different frequency planning techniques

On the other hand, the overdesign on reuse factor by the first technique can lead to higher user throughput. In Figure 26, it shows that the first technique can lead to much higher percentage of users served by 64QAM in the MR network. However, it is achieved by the massive expense on system capacity.

Figure 26. MCS percentage in access links

5.2 A New Measurement and Reporting Mechanism for