2.4 Joint Uplink and Downlink Green Network Coverage Problem
2.4.2 COMIC for Joint Uplink and Downlink Coverage Preservation 55
We extend the proposed COMIC algorithm for joint uplink and downlink network coverage preservation. For symmetric uplink and downlink, COMIC algorithm can be applied with Cm =Cmu T Cmd in which Cmu is the uplink coverage satisfying the uplink coverage requirements, Cmd is the downlink coverage satisfying the uplink coverage requirements.
To ensure the coverage of a network with both uplink coverage-limited BSs and downlink coverage-limited BSs, we propose the following asymmetric COMIC al-gorithm for the asymmetric uplink and downlink. Specifically, in the asymmetric COMIC algorithm, we first ensure the network uplink coverage and then minimize the network are power consumption for maintaining network downlink coverage. In this way, the extra power consumption to activate a BS for maintaining downlink
coverage will be smaller for a BS which is already activated for maintaining uplink coverage. With the preferences in already activated BSs in asymmetric COMIC, more constant operational power of BSs can be saved.
The steps of the asymmetric COMIC algorithm are as follows.
1. Choose a BS which has the maximal coverage range from B (say B1) as the initial active BS, and add it as the first element of the activated uplink BS set, denoted by Bonu , (i.e., Bonu ={B1}).
2. Find Bk which has the maximal coverage range from B \ {B1} and Cku∩ C1u 6= ∅.
Set Bk as the new active BS. Add the new active BS Bk toBon. Denote the set of uncovered boundary intersections between activated BSs as Iu and initiate Iu =∅.
3. For each Bm ∈ Bonu , find boundary intersection i between Bk and Bm which is not covered by any BS in the set of Buon\ {Bm}. Add boundary intersection i toI. Repeat until no such intersections exist for Bm, i.e.,
Iu =Iu[
{i|i ∈ \
Bj∈{Bonu \{Bm}}
Cj′
\∂Cmu
\∂Cku
\A},
where Cj′ represents the complement of Cju, ∂Cmu is the boundary of Cmu, and A is the target covering region.
4. For each new i ∈ Iu, find the optimal BS for i. The optimal BS for i is the BS which is closest to the optimal BS position of i and can cover the boundary intersection i. Set these BSs as candidate BSs.
5. Choose Bk which minimizes the distance from its position to the optimal posi-tion of the corresponding boundary intersecposi-tion from the candidate BSs as the new active BS. Add the new active BS Bk to Bonu .
6. For each i ∈ Iu, if i is covered by Bk, remove it fromIu. 7. Go to Step 3 until Iu =∅.
8. Find uncovered boundary intersections at network boundaries, i.e.,
I∂Au = [
Bm∈Buon
∂Cmu ∩ Con′ ∩ ∂A (2.57)
where I∂Au is the set of uncovered boundary intersections between active cells and the target area A, Conu =S
Bm∈BuonCmu and Con′ is the complement of Conu . 9. If I∂Au 6= ∅, choose Bk such that Cku minimize the overlap with Conu and can
cover a boundary intersection j ∈ I∂Au . For each j ∈ I∂Au , if j is covered by Bk, remove it from I∂Au . Add the new active BS Bk to Bonu .
10. Go to Step 3 until I∂Au =∅.
11. For the target area power consumption ζ, find the maximal downlink coverage for each BS {C1d, . . . ,CMd }.
12. Choose a BS which has maximal downlink coverage range from B (say B1) as the initial active BS, and add it as the first element of the activated downlink BS set, denoted by Bdon, (i.e., Bdon={B1}).
13. Find Bkwhich has maximal uplink coverage range fromB\{B1} and Ckd∩C1d 6= ∅.
Set Bkas the new active BS. Denote the set of uncovered boundary intersections between activated BSs as Id and initiate Id=∅.
14. For each Bm ∈ Bond , find boundary intersection i between Bk and Bm which is not covered by any BS in the set of Bdon\ {Bm}. Add boundary intersection i toId. Repeat until no such intersections exist for Bm, i.e.,
Id =Id[
{i|i ∈ \
Bj∈{Bdon\{Bm}}
Cj′′
\∂Cmd
\∂Ckd
\A},
where Cj′′ represents the complement of Cjd and ∂Cmd is the boundary of Cmd. Finally, add the new active BS Bk to Bond .
15. For each new i ∈ Id, find the optimal BS for i. The optimal BS for i is the BS which is closest to the optimal BS position of i and can cover the boundary intersection i. Set these BSs as candidate BSs. If no BS can cover the boundary intersection i, set ζ = ζ + ∆ζ and go to Step 11.
16. Choose Bkwhich minimizes the distance from its position to its optimal position of the corresponding boundary intersection from the candidate BSs as the new active BS.
17. For each i∈ Id, if i is covered by Bk, remove it from Id. 18. Go to Step 11 until Id=∅.
19. Find uncovered boundary intersections at network boundaries, i.e.,
I∂Ad = [
Bm∈Bdon
∂Cmd ∩ Con′′ ∩ ∂A (2.58)
where I∂Ad is the set of uncovered boundary intersections between active cells and the target area A, Cond =S
Bm∈BdonCmd and Con′′ is the complement of Cond . 20. IfI∂Ad 6= ∅, choose Bksuch that Ckdminimize the overlap with Cond and can cover
a boundary intersection j ∈ I∂Ad . For each j ∈ I∂Ad , if j is covered by Bk, remove it from I∂Ad . Add the new active BS Bk to Bond .
21. Go to Step 11 until I∂Ad =∅.
2.4.3 Performance Evaluation for Various Network Struc-tures
In this section, we evaluate the performance of various potential network architectures in HetNets for energy saving during the low traffic load period. In the simulations, we consider a network with regularly deployed macro BSs and randomly deployed small cell BSs. Specifically, we will compare the minimal network power consumption
Table 2.2: Simulation parameters of COMIC for joint uplink and downlink.
Parameter Value
channel model path loss+shadowing
shadowing standard deviation σΨ 6 dB minimum received power of macro BS -120 dBm minimum received power of small cell BS -108 dBm minimum received power of MS -100 dBm
antenna gain of macro BS 15 dBi
antenna gain of small cell BS 2 dBi transmit power efficiency of macro BS 0.5 transmit power efficiency of small cell BS 0.5
outage probability pout 0.1
path loss exponent α 4
network size 30 Km by 30 Km
K (reference distance = 1 m) for macro BS 10−1 ratio of uplink constant operational power 0.3 ratio of downlink constant operational power 0.3 ratio of active constant operational power 0.4
among only power control, cell activation with symmetric uplink and downlink, and cell activation with asymmetric uplink and downlink in HetNets. Here we compare only the network power consumption to avoid signal coverage holes during the low traffic load periods. The simulation parameters are given in Table 2.2.
Figure 2-9 and Figure 2-10 show the aggregated power consumption of BSs and the corresponding lower bounds for grid BS deployment and hexagonal BS deployment in a network, respectively. In the figures, compared with only power control, we can observe that the gain of cell activation with symmetric connection depends highly on the network topologies. The reason is that the uplink coverage restricted by MS maximum transmit power will limit the opportunities for a cell to extend its coverage with the requirements on symmetric uplink and downlink connections between MSs and BSs. By decoupling uplink connection and downlink connection by the proposed reception mode design for BSs, we demonstrate that the network power consumption can be further reduced. In the figure, we can observe that 1-2 dB gain can be obtained by asymmetric uplink and downlink.
Besides, with the exploration of data rate demands in the future, small cells
!
"#!
"$!
"%!
%&!
%'!
%#!
%$!
%%!
(&!
)*+,-!.*/0-*1! 2344,0-5.!67897! :2344,0-5.!678!97!
!"#$%&'()%$"&(*%!+,-)#.%!(/01-2
254;1:05*/!
1*+,-!<*;/=!
Figure 2-9: The network power consumption of grid deployment of 900 BSs for different power saving strateiges. Both the simulation results and the lower bounds of network power consumption are shown.
are expected to be densely deployed in the network for bursting network capacity.
An interesting question is whether network densification by small cells for bursting network capacity can further reduce network energy consumption during the low traffic load period. In the following, we evaluate various potential network structures in HetNets for network energy saving.
Figure 2-11 and Figure 2-12 show the network power consumption as a function of the constant operational power of a small cell with various network structures. Taking a glance in the figures, we can observe that not all schemes with small cells can achieve a better performance than activating only macro BSs. Compared with asymmetric uplink and downlink connections by only macro BSs, a promising approach which can achieve lower network power consumption with 4000 small cells in the network is the approach of uplink by small cells and downlink by macro cells. From Figure 2-11, we can find that uplink by macro BSs and downlink by small cells may be our least choice for network energy saving. Due to the fact that uplink coverage of a small cell is similar to uplink coverage of a macro cell, more network energy can be conserved
!
$>!
"&!
">!
%&!
%>!
(&!
)*+,-!.*/0-*1! 2344,0-5.!67897! :2344,0-5.!678!97!
!"#$%&'()%$"&(*%!+,-)#.%!(/01-2
254;1:05*/!
1*+,-!<*;/=!
Figure 2-10: The network power consumption of hexagonal deployment of 368 BSs for different power saving strateiges. Both the simulation results and the lower bounds of network power consumption are shown.
by activating a small cell which has a much lower uplink power consumption than that of a macro BS. On the other hand, due to the better transmit power efficiency and antenna gain of macro cells, the approach of downlink by macro BSs will be more energy-efficient than the approach of downlink by small cells. Therefore, the approach that uplink by small cells and downlink by macro cells has a much better performance than other approaches. From Figure 2-11, we can observe that this approach is better than all the other approaches as long as the constant operational power of a small cell is sufficiently small. Compared with network coverage preservation by only macro BSs, around 6 dB more power can be conserved by the approach (uplink by small cells and downlink by macro cells). To evaluate the network energy saving by network densifcation with small cells, we also show the results for a large amount of small cells deployed in the network. In this case, the increase of the deployed small cells will hardly affect the network power consumptions. Figure 2-12 shows the network power consumption as a function of the constant operational power of a small cell. In Figure 2-12, we can find that the network power consumption will be significantly reduced
!
#&! #>! >&! >>! $&! $>!
!"#%$&'()%$"&(*%!+,-)#.%!(/01-2
Figure 2-11: The network power consumption as a function of the constant operational power of a small cell. Various network structures in HeNets are examined with 368 hexag-onally deployed macro-cell BSs and 4000 randomly deployed small-cell BSs.
in the approach of maintaining coverage by only small cells compared with only 4000 small cells deployed in the network. With the state of art of power consumption in small cells (e.g., at the level of 10 Watts), maintaining network coverage by only small cells seems not the most energy-efficient approach. The reason is that the antenna gain provided by macro cells (e.g., by sectorization) can greatly reduce the transmit power consumption for maintaining downlink coverage. The lower gain omni-directional antenna adopted by small cell results in more power consumption by maintaining downlink coverage. The approach that maintaining uplink network coverage by small cells and downlink network coverage by macro cells will be most energy-efficient. Besides, it will be fitted into the proposed structure that an umbrella cell and multiple small cell to form a two-tier network [1]. It is intuitive that the umbrella cell serves as a downlink coverage preservation cell and the other small cells serve as uplink coverage preservation cells in the two-tier networks. Furthermore, to reduce signalling latency, the umbrella cell can transmit the control signalling to the small cells through the downlink channels during the low traffic load period under
!
#&! #>! >&! >>! $&! $>!
!"#%$&'()%$"&(*%!+,-)#.%!(/01-2
Figure 2-12: The network power consumption as a function of the constant operational power of a small cell. Various network structures in HeNets are examined with 368 hexag-onally deployed macro-cell BSs and 40000 randomly deployed small-cell BSs.
the two-tier network structure with downlink by only umbrella cells.