Many methods have been proposed for load balance. There are two categories to achieve load balance: power control based and handover based. In the meantime, there are several ways to trigger load balance methods.
3.1 Categories to Achieve Load Balance
One of the categories to achieve load balance is to control the pilot power of BSs. In general, if the BS overloads, it decreases its pilot power in order to prevent incoming connections. The other category to achieve load balance is handover. When the BS overloads, it can force one of its serving UEs to handover to the neighbor BS, and loading of the BS can be released.
3.1.1 Power Control based Category
In [4], authors calculate the relative load factor of each BS initially. The pilot power is reduced if the relative load factor was more than the threshold, and the pilot power is increased if the relative load factor was less than the threshold. However, decreasing the pilot power causes blind spots which are defined as the position where Ec/N0 of the received pilot signal is lower than a predefined threshold. After the initial adjustment, authors execute the second step based on the blind spot ratio. Authors assume that mobile stations can report the quality of received pilot signal. And authors estimate the blind spot ratio according to the reported quality. In the second step, with the increase in blind spot ratio, the probability to strengthen the pilot signal increases. On the contrary, the lower the ratio, the higher the probability is to reduce the power of pilot signal. Finally, the proposed method adjusts the pilot power again according to the probability acquired in the second
12
step. Load balance through pilot power adjustment can control the number of connecting users efficiently. However, UEs that are in idle mode must receive the pilot signal periodically. Adjusting the power of pilot signal means that some UEs would not be capable of camping on BSs. So it is not practical.
In [5], each local coverage area is treated as a bubble. The air within each bubble can be analogous to the traffic served by each cell. Temporary vacuum is treated as an un-served traffic. Bubbles must oscillate to fulfill the vacuum, and it alters the size of each coverage area. Altering the size of coverage area means to control the power of pilot signal. Treating each coverage area as a bubble is novel idea, but the proposed method needs the angles between UEs and BSs, and it is very difficult to acquire those angles. So the proposed method is difficult to implement.
3.1.2 Handover based Category
Most of the balancing techniques utilize handover to achieve load balance. In [6], both WLAN APs or Universal Mobile Telecommunication Service (UMTS) BSs calculate their utilities and broadcast it. Each mobile station receives the broadcasted utilities. If the utility of the serving AP (BS) is less than the received utility, the mobile station handovers to the network that has a higher utility. By introducing network utility, the proposed method is capable of being used in WLAN/UMTS interworking system. However, simulation scenarios seem to be having great impact on the performance of the proposed method, so it may be not suitable to deploy the proposed method in the real world.
In [7], there is an entity called AP Resource Advertisement Server (ARAS). Firstly, ARAS collects the bandwidth information of each UE. Secondly, ARAS calculates the available bandwidth of each AP based on the collected information, and then ARAS transmits the available bandwidth information to each “AP to AP Side Advertisement
13
Translator” (ASAT). And ASAT forwards that information to UEs by multicast. Finally, UEs periodically select the best AP by the received available bandwidth information. Obviously, it is clever to transmit the bandwidth information using multicast because it can reduce message traffic. However, there is a drawback if all the UEs receive the same bandwidth information. The AP with maximum available bandwidth would be selected by all the UEs that receive the same message, and then the AP would be congested.
3.2 Trigger Events for Load Balance
Load balance methods can be triggered by several events. One of the trigger events is that cell overloads. Obviously, it’s very reasonable to perform load balance method when cell overloads. Another trigger event is that the balance index among cells is less than the predefined threshold, and the balance index is calculated by the loading of those cells. If loading of those cells are the same, the balance index is 1. On the other hand, the more uneven the loading distributes, the lower the index is.
3.2.1 Overload based Event
proposed balance method only stops when the acquired value of Eq. (1) is less than zero.The proposed method does not mention how to select the UE for handover if there are multiple UEs that all of them can make Eq. (1) zero. In reality, users would move around.
UE handovered by load balance may be handovered back because of the movement of the UE, so the proposed method seems to be inappropriate if all the UEs are movable, and this
14
is why selecting the UE for handover from multiple UEs is so important.
3.2.2 Balance Index based Event
In [9], the balance index β is calculated first, and it is defined as Eq. (2).
k
i i
k
i i
k 1
2 2
1 )
(
(2)
If β is less than 1, the proposed method in [9] is executed. The method sorts those cells into three groups: under-loaded, balanced and overloaded. Those under-loaded cells allow new or handover connections. Balanced cells allow new connections only. And over-loaded cells deny any new or handover connections. And the proposed method selects the candidate mobile station for load transfer. The transferred mobile station is handovered from the overloaded cell to the under-loaded or balanced cell. Again, the proposed method does not consider the scenario that UEs are movable. When overloaded, the proposed method may choose the UE that would handover back to the source cell.
15