Escolar Documentos
Profissional Documentos
Cultura Documentos
Marco CARETTI
Virtual RAN
Conclusions
iJOIN Winter School
Marco Caretti – Wireless Innovation 2
Telecom Italia group 1/2
5G
Drivers
INCREMENT OF
THROUGHPUT IMPROVEMENT OF
COVERAGE
Peak (100 Mbps
Gbps) INCREMENT OF REDUCTION of
Service continuity in all the
and cell edge CAPACITY FOR A END-to-END sites of interest
GIVEN BANDWIDTH LATENCY (both outdoor and indoor)
More users and more Huge impacts on
traffic by managing QoS/QoE for 5G
the complexity services
…all with an increasing attention to Energy Efficiency and sustainability of evolved mobile networks,
driving to new network architectures
iJOIN Winter School
Marco Caretti – Wireless Innovation 6
RAN architecture evolution towards C-RAN / vRAN
The evolution towards the C-RAN / vRAN can be roughly summarized
in the following phases:
Phase 0: distributed BBU:
In most of current deployments the BBU is distributed in each site
There coud be a local connection (usually with CPRI) between BBU and RRU connected to the
antennas with coaxial cables (non split architectures)
The RRU are close to the antennas and connected with fibers to the BBU in the site
Loose coordination via X2 in case of LTE system
Energy
stations or
multi-RAT
installations
could
require still
space in
the site Compact solutions are already
installed with very low footprint
From [2]
In case of virtual RAN the SDR will further make easier the management enabling the
RAN upgrade by simple SW upgrade.
Moreover virtual RAN architecture will potentially allow lower HW/SW cost for the
operator due to the increased number of vendors (coming from the IT industry) and
further market segmentation and competition.
Efficient inter-RAT traffic management (e.g. inter-RAT mobility and load management).
Other functionalities (like ICIC, feICIC, NAICS), even if designed to support inter-eNB
communication through standardized interfaces (i.e. X2 interface) can exploit centralized
configurations.
Better HO performances
can be obtained in CRAN
architectures for some
scenarios like high
speed trains
From [1]
More efficient solutions, based for example on functional split between BB and RRU
are needed in the long term
considering further evolution of the
standard (higher order MIMO,
carrier aggregation with more than
2 carriers) multi-sectors/multi-RAT From [1]
configurations and HetNet
deployments
Up to now the RRU are P2P connected with BBU by means a fibre pair each. In a CRAN
architecture with high number of sites connected
to the BBU, if no more efficient solutions are
available (thanks to compressed solutions(*)
or different functional splits), additional transport solutions should be considered
Passive or active WDM
Using higher CPRI line bit rate, up to 10 Gb/s or more and electrical multiplexing (daisy chain)
(*) compression ratios are limited to 1/2 or 1/3 with negligible performance
loss iJOIN Winter School
Marco Caretti – Wireless Innovation 16
[1] C-RAN: The Road Towards Green RAN, China Mobile Research
Institute.
CRAN fronthauling transport challenges (2/2)
Passive WDM is the main available
technology used to multiply several
optical connections over one fiber
(pair), Typical values for the traditional
passive CWDM
Up to tens CPRI links over a single fiber
Passive – no power
Coloured SFP on both RRU and BBU sides
Lower cost
Active WDM
Requires power
Allow a better network monitoring
Grey SFP on both RRU and BBU sides
Both Ring or Bus topology
Probably Vendor dependent
More expensive
E-UTRAN
eNB
RAP
RAP
UE
Virtual
BaseBand pool
RAP
C-RAN architecture offers to operator the possibility to use green and efficient
infrastructure in order to save costs.
Open interfaces between the modules increase the possibility for algorithms
customization (for example in case of SON, Self Organizing Network algorithms).
Moreover, the path toward RAN virtualization will open new opportunities: HW/SW
decoupling, multivendor I/O, flexible deployments, …
In addition, new 5G services will introduce more challenging QoS requirements
(especially in terms of latency, energy efficiency, …) that may be difficult to fulfill
only with the implementation of C-RAN.
Then, in the view of future 5G systems, a significant advancement w.r.t. C-RAN
can be provided by the introduction of mobile edge computing:
moving cloud capabilities towards the network edge (i.e. eNBs)
and providing the system architecture with a joint management of
communication and computation resources.
iJOIN Winter School
Marco Caretti – Wireless Innovation 21
Network evolution toward 5G
The need for Mobile edge computing
Mobile edge computing (MEC) permits to extend cloud platform for applications to
the edge of the network (base stations), in order to improve the user experience
(also thanks to the standardization of proper measurements and network APIs).
E-UTRAN
EPC
eNB
+ MEC
PCRF
server HSS
Application
Server
RAP
RAP
This view is not in contrast with C-RAN architecture: MEC will add flexible
decentralization and proper dynamic instantiation and orchestration of virtual
machines serving for network management in close proximity to terminals (that are
also empowered with additional capabilities through computation offloading).
iJOIN Winter School
Marco Caretti – Wireless Innovation 22
Standardization activities relevant for MEC
Focus on ETSI
The final step of CRAN architectures is the virtual RAN architecture where a
reconfigurable HW (e.g. general purpose) is used with a significant number of
controlled cells, and BB resources in principle can be located in the same or
different physical locations (the cloud).
The processing power available at the edge of the network and possible new
requirements enable the addition of mobile edge computing on top of network
virtualization framework creating new opportunities for OTT and operators.
Marco Caretti
marco.caretti@telecomitalia.it