Proposed algorithm

Under the multi-femtocell environment, it seems that power control method for saving users might have some limitation. The orthogonal channel allocation is still the most efficient way to avoid strong interference and save the outage users. How to assign the orthogonal subchannels with other femtocells is an important issue.

According to the modulation coding scheme (MCS) table, listing the minimum SINR requirement of each MCS, it shows that the higher SINR has less difference between threshold values in dB. This indicates that reducing the serving power of users with good channel conditions to mitigate interference for users with bad channel conditions doesn‟t seem like a cost-effective way in system throughput view, unless the femtocell is the source of multiple users‟ maximum interference. This conclusion of observation is like the concept of waterfilling power control which says allocating more power to good channels increases the total capacity.

This method uses orthogonal subchannel allocation for poor users to increase the

15

system throughput and also takes care of the QoS requirement of all users. Because orthogonal subchannel allocation requires co-work between femtocells, a trading rule designed in proposed method defines the interaction of femtocells for fairness.

Besides, for the preservation of QoS, users‟ conditions about channels and data traffic are considered. The operation of proposed method is separated into three steps. The first step is calculation of subchannel number for each user. Second step assigns candidate subchannels to users. At last the third step classifies the conditions for orthogonal subchannel allocation and does subchannel selection. Because the all process are built on the concept of trading rule, following starts with the description and explanation of trading rule, and then introduces the steps of proposed method.

Trading rule is a protocol designed for ruling femtocells exchanging subchannels in proposed method. Orthogonal subchannel partition requires consideration about UEs‟ channel conditions and femtocell cooperation. If a UE asks for using orthogonal subchannels, it not only requires its serving femtocell to allocate UE‟s data transmitted on the particular subchannels, but also requires the interfering femtocells not to use those subchannels. The request helps the femtocells to save poor users but also reduces other femtocells‟ radio resource, so the rule is required for the fairness between femtocells, otherwise every femtocell just sends more request to get better performance of itself.

Trading rule says that a femtocell can require an orthogonal subchannels only by providing an unused subchannel. The setting of exchanging same number of subchannels is for fairness. The requirement exists depending on the comparison of SIR values and “interfered threshold”. UE calculates each SIR values by dividing received serving femtocell power with each received interfering femtocell power. SIR value lower than the interfered threshold means the interfering signal strength is above the acceptable level and the interfered femtocell will transmit orthogonal subchannel allocation request to the interfering femtocell for mitigate interference.

The idea of SIR values and the interfered threshold is shown in Figure 10. The following steps of proposed method obey the trading rule.

Figure 10 Orthogonal channel allocation request.

16

In the first step of algorithm, the amount of radio resource assigned to a user is based on its traffic amount requirement and channel condition. All of users under the same femtocell are assigned a number of candidate channels which is calculated with traffic amount requirement divided by available data transmission rate in subchannels.

Because every UE needs to face the bargain, they should prepare double number of candidate subchannels. After the allocation for QoS requirement, the remaining subchannels are assigned with the ratio of UEs‟ data traffic demands for fairness.

Each subchannel is assigned the same power budget. Here the reason of allocating user resource budget first is to fulfill QoS requirement and fairness issues at this step.

The users should have different resource allocation arrangements based on their channel condition. After the first step of proposed method, every user has its own power budget and number of subchannels, and these radio resources should be utilized carefully. For the users significantly interfered, the first priority should be increasing its channel quality, which also means orthogonal subchannel allocation by using half number of candidate subchannels for data transmission and the other half number of candidate subchannels for exchange the orthogonal subchannels with other femtocells, because using all the candidate subchannels not only lets the channel quality below the transmission threshold but also brings unnecessary interference into the system, degrading the system performance. On the other hand, users with good channel condition should use all of candidate subchannels under the same limited number of subchannels to increase the system throughput.

With the knowledge of the previous paragraph, the main issue is discussed here:

Which users belong to different femtocells should be allocated to the same candidate subchannels? Now considering a two-pair transmission scenario in the same group of subchannels with interference to each other, there are a few kinds of situation discussed as following.

First situation is that both of users have good channel conditions, so it has no need to use separated subchannels because sharing the spectrum by co-channel transmission is more efficient than by orthogonal channel transmission. Under this case, using the orthogonal subchannels for transmission wastes half of subchannels.

Although the channel quality of those orthogonal subchannels is better, the spectrum efficiency gain by SINR increment is hardly comparable to the subchannel resource wastage.

The second situation is that one of users has channel condition below the minimum data transmission threshold. If the user is not far below the threshold, it might be saved by power reduction of the other femtocell. But the method, lowering

17

signal power of good users to save poor users, is not cost-efficient that mentioned before, and it is not fair from the femtocell point of view. In addition, the case of users which can be saved by power adjustment seldom exists. The orthogonal channel separation saves most of outage users because of interference elimination, but at meanwhile reduces the throughput of good users. This situation has uncertainty gain of system throughput.

The third case is both of users are in outage. Orthogonal channel allocation for this case is the best way and only way to rescue users from outage. The total system throughput is increased if at least one of them is saved. Power control without orthogonal channel is not effective for this situation. Even the poor channel condition is not caused by strong interfering but weak serving power, using orthogonal channel selection still helps recovering the data transmission.

For those reasons before, sorting the users by their channel conditions and indicating the group of subchannels to users sequentially is a good solution. By this way, the candidate subchannels of poor users in different femtocells are allocated at the same bandwidth, so are candidate subchannels of good users shown as Figure 11.

The poor users use orthogonal subchannel allocation to save each other, and good users use co-channel transmission to increase the number of transmission subchannels for system throughput improvement. So in step 2, each femtocell sorts its users and assigns the candidate subchannels to them following the order of logical subchannel index.

Figure 11 Assign users subchannels by channel condition

In the last step, the femtocells exchange the control message to do the orthogonal subchannel selection. Femtocells transmit SIR values and index of subchannels by backhaul connection to those nearby femtocells which cause the SIR values below the threshold. After this process, every femtocell knows the channel condition interfered

18

by other femtocells and other cells‟ channel condition interfered by it. After that, the femtocells do resource allocation one by one in a random order mentioned in assumption. For those users with SIR values below the threshold, they are required to do orthogonal subchannel partition. Here includes two kinds of subchannel selection modes. One is concerning strongest interfering or interfered source, and the other one is counting the interfering or interfered times. The first kind of selection mode only considers the channel selection result of strongest interfering femtocell, and the interfered femtocell chooses the subchannels that are not occupied by the strongest interference. The second mode needs to calculate the occupation frequency of interfered and interfering femtocells on each subchannels. Femtocells choose the subchannels with minimum value of occupation frequency for reducing interference.

These two modes have small difference but should have similar performance under proper “interfered threshold” condition. This step also includes two power control modes that one is no power adjustment and the other is allocating remaining power on those orthogonal subchannels for improving channel conditions and system throughput. These two power control modes affect the system power consumption.

Figure 12 gives an example. After the resource allocation, femtocells need to broadcast its control information to other femtocells to let them know.

Figure 12 Two subchannel selection mode and two power adjustment mode

This method requires 4 parts of information exchange. At first, femtocells send control signal like preamble or pilot signal to let users measure the channel condition and SIR values. Secondly, UEs send the SIR values with femtocell indexes which below the SIR threshold to its serving femtocell. Third, femtocells exchange the SIR values and indexes of subchannels by backhaul to aware the nearby cells. At last, each femtocell transmits its control decision to warn other femtocells which haven‟t do resource allocation yet. The information exchanges in proposed algorithm don‟t cost too much radio resource because most information goes by backhaul, and the amount is small. Whole process is showed in Figure 13.

19

Figure 13 Procedure of information exchange.

The control flow of each femtocell based on proposed method is like Figure 14.

Figure 14 Control flow of each femtocell

20