Performance analysis and implementation of modified sdm based noc for mpsoc o...
LTE_Cloud_RAN_MAC_HLD
1. 1Gbps CLOUD RAN MediumAccessControl Sub-System
SOFTWARE ARCHITECTURE FRAMEWORK FOR Data-Plane of Long Term
Evolution(LTE) eNodeB WITH 1Gbps CLOUD RAN MEDIUM ACCESS CONTROL
(MAC) SUB-SYSTEM
High Level Design Document
1 Abstract
This document gives the summary of the next generation software architecture framework of Long
Term Evolution (LTE) eNodeB data-plane with Cloud RAN Medium Access Control (MAC) sub-system.
This document also addresses the data – throughput bottleneck which a user experiences in the
currentlydeployedeNodeB(s),whichistargetedataround1 gigabits/sec.
Keywords–LTE CloudRAN,4G-LTE 1 Gbps MAC framework.
2 Pre-requisites
To understand the high level design of the below mentioned software architecture, first, the
knowledge of the current data-plane MAC sub-system for an eNodeB of Freescale Semiconductors is
required.
3 eMACs for an UE per eNodeB
From the Block Diagram of eMACs for an UE per eNodeB,it isevident that the framework mentioned
in Figure 1, gives the idea of spreading the operations of the eMACs not only to Time and Frequency
domains,butalsointhe third dimensionknow asthe Spatial domain.
Each MAC entity has got the downlink and uplink entities for a particular UE. For further in-depth
explanation of the MAC entities, the reader is required to understand Freescale Semiconductor’s LTE
eNodeBMACarchitecture.
Figure 1. BlockDiagram of LTE Downlink eMACsforanUE per eNodeB
0
1
2
n-
1
Internal UE Context Indices
0th
index to (n-1) index; where n is a power of 2
value
0 1 2 3 4 5 6 7 8 9
10 eMACs per UE
0 1 2 3 4 5 6 7 8 9
0 1 2 3 4 5 6 7 8 9
UEInternal Context Index0 mappedto 10eMACs
UEInternal Context Index1 mapped to 10eMACs
UEInternal Context Index2 mappedto 10eMACs
UEInternal Context Index(n-1) mapped to10eMACs
0 1 2 3 4 5 6 7 8 9
2. 1Gbps CLOUD RAN MediumAccessControl Sub-System
4 External C-RNTI to Internal UE Index mapping
Note:Where ‘N’isany RNTIwithinthe range of 0 to 65535.
Figure 2. Blockdiagramof External C-RNTItoInternal UE Index mapping
In the above mentioned Figure 2, the block diagram of a particular C-RNTI mapped to a particular
internal UE index of the eMACisprojected.
5 Downlink Hybrid Automatic Repeat request (HARQ) re-
transmissionmemory model
In the below mentioned Figure 3, the block diagram of the HARQ re-transmission memory model
associatedwiththe variouskindsof UE(s) isshownbelow.
0
1
2
n-
1
1
0
1023
512
2
10
2
65535
65534
N
1
0
External UEC-RNTI Indices
0th index to 65535index
Internal UE Context Indices
0th
index to (n-1) index; where n is a power of 2
value; maximum value for ‘n’ is 65536
3. 1Gbps CLOUD RAN MediumAccessControl Sub-System
HARQ re-transmission queuehead
UE-ID #0 ;
save the pointer to
the HARQ –entity
#0 and pointer to
HARQ – entity #9
HARQ
entity#4
HARQ
process #0
HARQ
process #N
HARQ
process #7
NULL NULL
HARQ
entity#5
HARQ
process #0
HARQ
process #N
HARQ
process #7
NULL NULL
HARQ
entity#6
HARQ
process #0
HARQ
process #N
HARQ
process #7
NULL NULL
HARQ
entity#7
HARQ
process #0
HARQ
process #N
HARQ
process #7
NULL NULL
HARQ
entity#2
HARQ
process #0
HARQ
process #N
HARQ
process #7
NULL NULL
HARQ
entity#3
HARQ
process #0
HARQ
process #N
HARQ
process #7
NULL NULL
HARQ
entity#1
HARQ
process #0
HARQ
process #N
HARQ
process #7
NULL NULL
HARQ
entity#0
HARQ
process #0
HARQ
process #N
HARQ
process #7
NULL NULL
HARQ
entity#8
HARQ
entity#9
NULL
NULL
NULL
A B
NULL
entity#0
NULL
entity
#0
4. 1Gbps CLOUD RAN MediumAccessControl Sub-System
Note: Where ‘X’ is any value from 0 to “Maximum number of internal UE – indices which does not
increase more than inclusive 65536” for MAC.
Note:Where ‘N’isany value inthe range of “0 to7” for FDD MAC.
Figure 3. Blockdiagramof the DL - HARQre-Transmission MemoryModel
The HARQ memorymodel supportsa3-Tiersearch.
First, it checks the availability of the UE entity, then secondly, it searches the HARQ eMAC entity
database and, finally, it checks the availability of the HARQ process node in the re-tranmission
queue.
6 Intended Audience
4G-LTE wirelessembeddedsoftware engineersand software architects.
HARQ
entity#0
HARQ
process #0
HARQ
process #N
HARQ
process #7
NULL NULL
HARQ
entity#3
HARQ
process #0
HARQ
process #N
HARQ
process #7
NULL NULL
HARQ
entity#6
HARQ
process #0
HARQ
process #N
HARQ
process #7
NULL NULL
HARQ
entity#8
HARQ
process #0
HARQ
process #N
HARQ
process #7
NULL NULL
A B
UE-ID # X;
save the pointer to
the HARQ –entity
#0 and pointer to
the last HARQ -
entity # 8
NULL
entity
#0
5. 1Gbps CLOUD RAN MediumAccessControl Sub-System
7 Platform Statistics
The Processor used is INTEL P2020 – 2.9 GHz with Microsoft Visual Studio C++ version 6 on Windows
7 – Professional (2009edition)
It is evident from the performance that about 40 % of the CPU Usage on an average is occupied by
the software framework.
In the demonstration,amaximumof 5UE(s) have beenadmittedtothe system.
8 Author
SubhankarSen