For the convenience of the comparison between HetNet andHomoNet or among different HetNet deployments, the metricof “cell group average throughput” is introduced to evaluate thesystem throughput per unit area, and the unit area is defined asthe single-macro-cell coverage area in the traditional macro-cell-HomoNet. The definition of “cell edge user throughput” isthe same as the traditional macro-cell-HomoNet, i.e. expressingthe 5th percentile of the worst user experience in the wholenetwork deployment.III.
ETNET FREQUENCY ALLOCATIONS
The frequency allocation among macro cells and local cellsare crucial to HetNet. The following three approaches areanalyzed from the aspects of the capacity and coverage:
Co-channel allocation: macro cells as well as local cellsshare the same entire frequency band.
Orthogonal frequency allocation: macro cells are assignedto a part of the whole frequency resource, and local cellwould use the remaining part, thus, frequency resource between macro cell and local cell is not overlapped.
Overlapped frequency allocation: Frequency resource between marco cells and local cells is overlapped partially.For example, Marco cells are assigned to a part of thewhole frequency resource, while the local cells occupythe whole frequency resource, and vice versa.
The corresponding examples of the three frequency allocationapproaches with 10MHz whole bandwidth are shown inFigure 1.
Figure 1. HetNet channel deployments
System capacity analysis
In this subSection the DL data channel system capacity of the three frequency allocation approaches in Figure 1 areevaluated with HetNet configuration 1, with the full buffer traffic model, 500m inter-site distance, 3GPP Urban channelmodel Case1 , and the proportional fair scheduler. TheHetNet system capacity gain over HomoNet for both cell groupaverage throughput and cell edge throughput are shown inFigure 2, where cell selection is based on the traditional RRMmeasurement of received power and quality of referencesignals. It can be seen that:
Frequency overlapped allocation (macro cell with 5MHzand pico cell with 10MHz) could reach a best cell groupaverage throughput due to the local-cell-friendly resourceallocation, but the cell edge throughput is deterioratedobviously over co-channel case since all the cell edgeUEs are macro cell edge UEs, which only have partial bandwidth access.
Orthogonal channel allocation has the worst cell groupaverage throughput due to half of the whole bandwidthcan be used for both macro cells and local cells network wide, but better cell edge throughput than the frequencyoverlap case because there is no inter-cell interference between macro cells and local cells.
Co-channel frequency allocation is a good balance between the cell group average throughput and cell edgethroughput, one reason is that both macro cells and localcells can make full use of the whole bandwidth, another reason is that the cell-edge UEs’ throughput areguaranteed with the proportional fair scheduler.It should be noted that above is only for data channel capacityevaluation, without taken the control channel coverage andquality into account.
Figure 2. HetNet system capacity gain over HomoNet, comparison amongthree frequency allocation approaches
Control channel coverage analysis
The system capacity evaluation in the last sub-Sectionassumes full buffer traffic without considering the controlchannel coverage. Local cell deployment, as an efficientmeasure to offload the hotspot UEs and high traffic from macrocells, would have different load and coverage depending on thefrequency planning.
Figure 3. SINR distributions of HetNet channel deployments (macro cell Tx power 46dBm)
Figure 3 shows the geometry contours of UEs among thewhole area with different local cell transmission powers for both co-channel and orthogonal frequency allocationapproaches, by calculating the average received signal tointerference and noise ratio (SINR) experienced by thereference signals and control channel. Users select macro cellor local cell with the highest received SINR, i.e. based on the