Variation in Blocking Probability with Different Aggregation Levels (ALs)
In this notebook, we simulate the
blocking probabilityas a function of differentALs.We consider 5 scenarios corresponding to each AL supported for USS. In each scenario only one of the AL is used the probability of 1.
For instance [1, 0, 0, 0, 0] distribution is used to evaluate the blocking probability when AL 1 is chosen.
Intuitively we see that a higher AL provides a better coverage but at the cost of using more CCEs and hence higher blocking probabilty.
We consider 5 scenarios corresponding to each Aggregation Level (AL) of interest and the distributions are as follows:
AL 1 distribution : [1, 0, 0, 0, 0]
AL 2 distribution : [0, 1, 0, 0, 0]
AL 4 distribution : [0, 0, 1, 0, 0]
AL 8 distribution : [0, 0, 0, 1, 0]
AL 16 distribution : [0, 0, 0, 0, 1]
The CORESET size
Nccepis 54 CCEs for this simulation.
Python Libraries
[1]:
import os
os.environ["CUDA_VISIBLE_DEVICES"] = "-1"
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'
# %matplotlib widget
import matplotlib.pyplot as plt
import matplotlib.patches as mpatches
import matplotlib as mpl
import numpy as np
5G-Toolkit Libraries
[2]:
import sys
sys.path.append("../../")
from toolkit5G.Scheduler import PDCCHScheduler
Simulation Parameters
The following parameters are used for this simulation: - coresetID denotes the coreset ID. - slotNumber denotes the slot-number carrying the PDCCH. - searchSpaceType denotes the search space type. UE specific search space (USS) or Common search space (CSS). - nci denotes the variable corresponding to carrier aggregation. Current simulation does not assume carrier aggregation.
[3]:
mu = np.random.randint(4) # numerlogy for sub-carrier spacing
numSlotsPerFrame = 2**mu * 10 # number of slots per radio frame
coresetID = 1 # coreset ID
slotNumber = 0 # slot number
searchSpaceType = "USS" # search space type. UE specific search space
nci = 0 # variable corresponding to carrier aggregation
numIterations = 1000 # Number of Iterations considered in simulation
PDCCH Scheduling Parameters
Following parameters are crucial for PDCCH scheduling performance: - Nccep denotes coreset size or number of CCEs available for scheduling UEs. - strategy denotes the scheduling strategy. - numCandidates denotes number of PDCCH candidates per each Aggregation Level.
[4]:
Nccep = 54 # coreset size in terms of CCEs
maxNumUEs = 60
numUEs = np.arange(0,maxNumUEs+1,4) # UE range considered in simulation
numUEs[0] = 1
numCandidates = np.array([6,6,4,2,1], dtype=int) # number of PDCCH candidates per AL
strategy = "Conservative" # scheduling strategy
pdcchSchedulerObj = PDCCHScheduler(mu,slotNumber,coresetID,nci)
Impact of AL 1
In this case the Base Station (BS) only configures 6 candidates (of AL 1) for each UE to monitor and other ALs are not monitored.
[5]:
########
# AL1
########
aggLevelProbDistributionForAL1 = np.array([1, 0.0, 0.0, 0.0, 0.0])
probOfBlockingForAL1 = np.zeros(numUEs.shape)
for n in range(numUEs.size):
print("Simulating (n,numUEs) : "+str(n)+", "+str(numUEs[n]))
prob = 0
for i in range(numIterations):
ueALdistribution = np.random.multinomial(numUEs[n], aggLevelProbDistributionForAL1)
rnti = np.random.choice( np.arange(1,65519+1), size = (numUEs[n],), replace=False)
count = pdcchSchedulerObj(Nccep,searchSpaceType,ueALdistribution,numCandidates,rnti,strategy)[0]
numBlockedUEs = np.sum(count)
prob = prob + numBlockedUEs/numUEs[n]
probOfBlockingForAL1[n] = prob/numIterations
Simulating (n,numUEs) : 0, 1
Simulating (n,numUEs) : 1, 4
Simulating (n,numUEs) : 2, 8
Simulating (n,numUEs) : 3, 12
Simulating (n,numUEs) : 4, 16
Simulating (n,numUEs) : 5, 20
Simulating (n,numUEs) : 6, 24
Simulating (n,numUEs) : 7, 28
Simulating (n,numUEs) : 8, 32
Simulating (n,numUEs) : 9, 36
Simulating (n,numUEs) : 10, 40
Simulating (n,numUEs) : 11, 44
Simulating (n,numUEs) : 12, 48
Simulating (n,numUEs) : 13, 52
Simulating (n,numUEs) : 14, 56
Simulating (n,numUEs) : 15, 60
Impact of AL 2
In this case the BS only configures 6 candidates (of AL 2) for each UE to monitor and other ALs are not monitored.
[6]:
########
# AL2
########
aggLevelProbDistributionForAL2 = np.array([0.0, 1, 0.0, 0.0, 0.0])
probOfBlockingForAL2 = np.zeros(numUEs.shape)
for n in range(numUEs.size):
print("Simulating (n,numUEs) : "+str(n)+", "+str(numUEs[n]))
prob = 0
for i in range(numIterations):
ueALdistribution = np.random.multinomial(numUEs[n], aggLevelProbDistributionForAL2)
rnti = np.random.choice( np.arange(1,65519+1), size = (numUEs[n],), replace=False)
count = pdcchSchedulerObj(Nccep,searchSpaceType,ueALdistribution,numCandidates,rnti,strategy)[0]
numBlockedUEs = np.sum(count)
prob = prob + numBlockedUEs/numUEs[n]
probOfBlockingForAL2[n] = prob/numIterations
Simulating (n,numUEs) : 0, 1
Simulating (n,numUEs) : 1, 4
Simulating (n,numUEs) : 2, 8
Simulating (n,numUEs) : 3, 12
Simulating (n,numUEs) : 4, 16
Simulating (n,numUEs) : 5, 20
Simulating (n,numUEs) : 6, 24
Simulating (n,numUEs) : 7, 28
Simulating (n,numUEs) : 8, 32
Simulating (n,numUEs) : 9, 36
Simulating (n,numUEs) : 10, 40
Simulating (n,numUEs) : 11, 44
Simulating (n,numUEs) : 12, 48
Simulating (n,numUEs) : 13, 52
Simulating (n,numUEs) : 14, 56
Simulating (n,numUEs) : 15, 60
Impact of AL 4
In this case the BS only configures 4 candidates (of AL 4) for each UE to monitor and other ALs are not monitored.
[7]:
########
# AL4
########
aggLevelProbDistributionForAL4 = np.array([0.0, 0.0, 1, 0.0, 0.0])
probOfBlockingForAL4 = np.zeros(numUEs.shape)
for n in range(numUEs.size):
print("Simulating (n,numUEs) : "+str(n)+", "+str(numUEs[n]))
prob = 0
for i in range(numIterations):
ueALdistribution = np.random.multinomial(numUEs[n], aggLevelProbDistributionForAL4)
rnti = np.random.choice( np.arange(1,65519+1), size = (numUEs[n],), replace=False)
count = pdcchSchedulerObj(Nccep,searchSpaceType,ueALdistribution,numCandidates,rnti,strategy)[0]
numBlockedUEs = np.sum(count)
prob = prob + numBlockedUEs/numUEs[n]
probOfBlockingForAL4[n] = prob/numIterations
Simulating (n,numUEs) : 0, 1
Simulating (n,numUEs) : 1, 4
Simulating (n,numUEs) : 2, 8
Simulating (n,numUEs) : 3, 12
Simulating (n,numUEs) : 4, 16
Simulating (n,numUEs) : 5, 20
Simulating (n,numUEs) : 6, 24
Simulating (n,numUEs) : 7, 28
Simulating (n,numUEs) : 8, 32
Simulating (n,numUEs) : 9, 36
Simulating (n,numUEs) : 10, 40
Simulating (n,numUEs) : 11, 44
Simulating (n,numUEs) : 12, 48
Simulating (n,numUEs) : 13, 52
Simulating (n,numUEs) : 14, 56
Simulating (n,numUEs) : 15, 60
Impact of AL 8
In this case the BS only configures 2 candidates (of AL 8) for each UE to monitor and other ALs are not monitored.
[8]:
########
# AL8
########
aggLevelProbDistributionForAL8 = np.array([0.0, 0.0, 0.0, 1, 0.0])
probOfBlockingForAL8 = np.zeros(numUEs.shape)
for n in range(numUEs.size):
print("Simulating (n,numUEs) : "+str(n)+", "+str(numUEs[n]))
prob = 0
for i in range(numIterations):
ueALdistribution = np.random.multinomial(numUEs[n], aggLevelProbDistributionForAL8)
rnti = np.random.choice( np.arange(1,65519+1), size = (numUEs[n],), replace=False)
count = pdcchSchedulerObj(Nccep,searchSpaceType,ueALdistribution,numCandidates,rnti,strategy)[0]
numBlockedUEs = np.sum(count)
prob = prob + numBlockedUEs/numUEs[n]
probOfBlockingForAL8[n] = prob/numIterations
Simulating (n,numUEs) : 0, 1
Simulating (n,numUEs) : 1, 4
Simulating (n,numUEs) : 2, 8
Simulating (n,numUEs) : 3, 12
Simulating (n,numUEs) : 4, 16
Simulating (n,numUEs) : 5, 20
Simulating (n,numUEs) : 6, 24
Simulating (n,numUEs) : 7, 28
Simulating (n,numUEs) : 8, 32
Simulating (n,numUEs) : 9, 36
Simulating (n,numUEs) : 10, 40
Simulating (n,numUEs) : 11, 44
Simulating (n,numUEs) : 12, 48
Simulating (n,numUEs) : 13, 52
Simulating (n,numUEs) : 14, 56
Simulating (n,numUEs) : 15, 60
Impact of AL 16
In this case the BS only configures 1 candidate (of AL 16) for each UE to monitor and other ALs are not monitored.
[9]:
########
# AL16
########
aggLevelProbDistributionForAL16 = np.array([0.0, 0.0, 0.0, 0.0, 1])
probOfBlockingForAL16 = np.zeros(numUEs.shape)
for n in range(numUEs.size):
print("Simulating (n,numUEs) : "+str(n)+", "+str(numUEs[n]))
prob = 0
for i in range(numIterations):
ueALdistribution = np.random.multinomial(numUEs[n], aggLevelProbDistributionForAL16)
rnti = np.random.choice( np.arange(1,65519+1), size = (numUEs[n],), replace=False)
count = pdcchSchedulerObj(Nccep,searchSpaceType,ueALdistribution,numCandidates,rnti,strategy)[0]
numBlockedUEs = np.sum(count)
prob = prob + numBlockedUEs/numUEs[n]
probOfBlockingForAL16[n] = prob/numIterations
Simulating (n,numUEs) : 0, 1
Simulating (n,numUEs) : 1, 4
Simulating (n,numUEs) : 2, 8
Simulating (n,numUEs) : 3, 12
Simulating (n,numUEs) : 4, 16
Simulating (n,numUEs) : 5, 20
Simulating (n,numUEs) : 6, 24
Simulating (n,numUEs) : 7, 28
Simulating (n,numUEs) : 8, 32
Simulating (n,numUEs) : 9, 36
Simulating (n,numUEs) : 10, 40
Simulating (n,numUEs) : 11, 44
Simulating (n,numUEs) : 12, 48
Simulating (n,numUEs) : 13, 52
Simulating (n,numUEs) : 14, 56
Simulating (n,numUEs) : 15, 60
Plot the Variation in Blocking Probability with number of UEs for different Aggregation levels.
Its the recreation of
Fig. 7: Blocking probability for different ALs (CORESET size = 54 CCEs)from the reference paper referenced below [1].From the figure we see that using a higher AL leads to higher blocking probability.
The results in figure shows that to maintain the blocking probability
below 0.2, the maximum possible number of UEs to be scheduled with ALs 2, 4, 8, and 16 is close to 33, 16, 6, and 2 respectively.
[10]:
fig, ax = plt.subplots()
ax.plot(numUEs, probOfBlockingForAL1, marker = "*", linestyle = "solid", ms = 9, c = 'b', label = "AL1")
ax.plot(numUEs, probOfBlockingForAL2, marker = "o", linestyle = "solid", ms = 6, c = 'y', label = "AL2")
ax.plot(numUEs, probOfBlockingForAL4, marker = "v", linestyle = "solid", ms = 6, c = 'k', label = "AL4")
ax.plot(numUEs, probOfBlockingForAL8, marker = "^", linestyle = "solid", ms = 6, c = 'g', label = "AL8")
ax.plot(numUEs, probOfBlockingForAL16, marker = "s", linestyle = "solid", ms = 6, c = 'm', label = "AL16")
ax.legend()
ax.set_xlabel('Total Number of Scheduled UEs')
ax.set_ylabel('Blocking Probability')
ax.set_title('Blocking Probability for different ALs', fontsize=12)
ax.set_xticks(numUEs[0:-1:2])
ax.set_yticks(np.linspace(0.05,0.95,10), minor=True)
ax.set_yticks(np.linspace(0, 1,11), minor=False)
ax.tick_params(axis='both',which='major', grid_linewidth=1, width=0)
ax.tick_params(axis='both',which='minor', grid_linewidth=0.5, grid_linestyle = '--')
ax.set_ylim([0, 1])
ax.set_xlim([0, numUEs[-1]])
ax.grid(which='both')
plt.show()
References
[1] Blocking Probability Analysis for 5G New Radio (NR) Physical Downlink Control Channel. Mohammad Mozaffari, Y.-P. Eric Wang, and Kittipong Kittichokechai
[ ]: