Other Inter-cell Interference Avoidance Schemes

Authors in [12] proposed Incremental Frequency Reuse (IFR) Scheme. Rather than SFR and FFR, IFR does not modify the frequency and power planning. It uses the same frequency and power planning as reuse-1 for all clusters. The whole bandwidth is partitioned into several

20

exclusive parts and the number of these parts is also the number of clusters in the system. The concept of IFR is illustrated in Figure 8. Each cluster has its own base segment, which means this segment must be allocated first. For reducing inter-cell interference, the segment allocation has specific order. We take Figure 8 as an example, the segment allocation sequences can be summarized as

 Sequence-A : Segment A → Segment B → Segment C

 Sequence-B : Segment B → Segment C → Segment A

 Sequence-C : Segment C → Segment A → Segment B

By applying these allocation sequences, UEs in different cells have less probability to use the same frequency band, hence, inter-cell interference can be eliminated. But interference can only be eliminated when system load is not so high. When UE is getting more, system load is close to full, performance of IFR is similar with reuse-1. At this situation, UE located at cell edge cannot access the system due to low cell coverage caused by severe inter-cell interference. From the discussion above we know that IFR is only suite for light loading system. Once system loading is increasing, performance will scale down severely.

Figure 8 - Concept of Incremental Frequency Reuse based on 3 partitioning segment.

21

All of the above schemes are static in frequency domain planning. In other words, these schemes cannot adjust the bandwidth of high power frequency band and normal power frequency band while system is running. Applying these methods may bring about the user distribution problem. If the predefined normal power frequency band is much larger than high

power frequency band and the user distribution is that cell edge users are much more than cell center user. Thus, some resources may be wasted because the cell edge user may not use the normal power frequency band due to the weak signal strength.

The user distribution problem is illustrated in Figure 9. The number of CEUs is more than the number of CCUs in cell A, cell B and cell C. Because the high power frequency band is too small to support CEU, some CEUs may not access the radio system. However, CCUs only occupy a little of normal power resources. There are still lots of resources remained idle.

But CEUs which cannot access the radio system cannot use the normal power resources due to the poor channel quality. Hence, according to the discussion above, dynamic interference coordination scheme may get better performance.

Figure 9 - If the a mount of cell edge user is large and the a mount of cell c enter user is sma ll, some ce ll edge users may not get enough resource. Although there are id le resources, cell edge user cannot access these resources due to the bad channel quality.

22

Here we introduce two dynamic schemes. The authors of [13] observe the performance of reuse-1 and reuse-3 with VoIP traffic and data traffic. It shows that the voice users are benefited by using reuse-3 and data traffic by using reuse-1. So a dynamic scheme named Adaptive Frequency Reuse is proposed. The concept is that each cell can dynamically adjust

the available frequency band according to the proportion of voice users. For example, if all users are not voice user, cells are allocated the entire bandwidth, what is equivalent to reuse-1 system. On the other hand, if the proportion of voice users in the system is one (all UEs are voice user), cells are allocated one third of the whole bandwidth, what is equivalent to a reuse-3 system. In this case, the way to allocate frequency band is like reuse-3, each cluster must uses exclusive frequency band. However, this scheme does not consider the frequency spectrum efficiency. In the system with a lot of voice user, the system capacity is pretty low because the available bandwidth a cell can use is close reuse-3.

Another distributed dynamic inter-cell coordination scheme which proposed in [14] can adjust the power and frequency band while system is running. The concept of [14] is to find the dominant interferer around a cell and send power restriction request through X2 interface to restrict the interference level. Although this method can get good performance, it suffers from complicated matrix calculation during finding the dominant interferer procedure. Also, each allocation run needs messages transmission a mong cells through X2 interface. It may cost too much time to finish the above job. The realistic environment constrain make this method not so practical.

To overcome the user distribution problem occurred at static scheme, the dynamic inter-cell interference coordination scheme is desirable. And for applying the practical system we need to design a scheme with the principle of low complexity and minimal messages exchange. The objective of desirable method is to retain the advantage of FFR which has the better performance both in system throughput and the cell coverage and also solve the user distribution problem. The new method must be designed to reach two main requirements:

23

 Support flexibility in frequency domain with non-uniform user distribution.

 Low system complexity and minimal messages exchange.

The first requirement is used to handle that when user distribution is various, each cell can re-configure to an appropriate frequency and power planning. The second requirement is used to suite on practical system architecture, because the high complexity algorithm with lots of complicated calculation and too many inter-cell communication messages between cells is not desirable.

24