Pasca, Thomas Valerrian and Tamma, Bheemarjuna Reddy
(2019)
Traffic Steering in Radio Level Integration of LTE
and Wi-Fi Networks.
PhD thesis, Indian institute of technology Hyderabad.
Abstract
A smartphone generates approximately 1, 614 MB of data per month which is 48 times
of the data generated by a typical basic-feature cell phone. Cisco forecasts that the mobile data traffic growth will remain to increase and reach 49 Exabytes per month by 2021.
However, the telecommunication service providers/operators face many challenges in order
to improve cellular network capacity to match these ever-increasing data demands due to
low, almost flat Average Revenue Per User (ARPU) and low Return on Investment (RoI).
Spectrum resource crunch and licensing requirement for operation in cellular bands further
complicate the procedure to support and manage the network.
In order to deal with the aforementioned challenges, one of the most vital solutions is
to leverage the integration benefits of cellular networks with unlicensed operation of Wi-Fi
networks. A closer level of cellular and Wi-Fi coupling/interworking improves Quality of
Service (QoS) by unified connection management to user devices (UEs). It also offloads
a significant portion of user traffic from cellular Base Station (BS) to Wi-Fi Access Point
(AP). In this thesis, we have considered the cellular network to be Long Term Evolution
(LTE) popularly known as 4G-LTE for interworking with Wi-Fi.
Third Generation Partnership Project (3GPP) defined various LTE and Wi-Fi interworking architectures from Rel-8 to Rel-11. Because of the limitations in these legacy LTE
Wi-Fi interworking solutions, 3GPP proposed Radio Level Integration (RLI) architectures
to enhance flow mobility and to react fast to channel dynamics. RLI node encompasses link
level connection between Small cell
deployments, (ii) Meeting Guaranteed Bit Rate (GBR) requirements of the users including
those experiencing poor Signal to Interference plus Noise Ratio (SINR), and (iii) Dynamic
steering of the flows across LTE and Wi-Fi links to maximize the system throughput.
The second important problem addressed is the uplink traffic steering. To enable efficient uplink traffic steering in LWIP system, in this thesis, Network Coordination Function
(NCF) is proposed. NCF is realized at the LWIP node by implementing various uplink traffic steering algorithms. NCF encompasses four different uplink traffic steering algorithms
for efficient utilization of Wi-Fi resources in LWIP system. NCF facilitates the network to
take intelligent decisions rather than individual UEs deciding to steer the uplink traffic onto
LTE link or Wi-Fi link. The NCF algorithms work by leveraging the availability of LTE as
the anchor to improvise the channel utilization of Wi-Fi.
The third most important problem is to enable packet level steering in LWIP. When
data rates of LTE and Wi-Fi links are incomparable, steering packets across the links create
problems for TCP traffic. When the packets are received Out-of-Order (OOO) at the TCP
receiver due to variation in delay experienced on each link, it leads to the generation of
DUPlicate ACKnowledgements (DUP-ACK). These unnecessary DUP-ACKs adversely affect the TCP congestion window growth and thereby lead to poor TCP performance. This
thesis addresses this problem by proposing a virtual congestion control mechanism (VIrtual
congeStion control wIth Boost acknowLedgEment -VISIBLE). The proposed mechanism
not only improves the throughput of a flow by reducing the number of unnecessary DUPACKs delivered to the TCP sender but also sends Boost ACKs in order to rapidly grow the
congestion window to reap in aggregation benefits of heterogeneous links.
The fourth problem considered is the placement of LWIP nodes. In this thesis, we have
addressed problems pertaining to the dense deployment of LWIP nodes. LWIP deployment
can be realized in colocated and non-colocated fashion. The placement of LWIP nodes is
done with the following objectives: (i) Minimizing the number of LWIP nodes deployed
without any coverage holes, (ii) Maximizing SINR in every sub-region of a building, and
(iii) Minimizing the energy spent by UEs and LWIP nodes.
Finally, prototypes of RLI architectures are presented (i.e., LWIP and LWA testbeds).
The prototypes are developed using open source LTE platform OpenAirInterface (OAI) and
commercial-off-the-shelf hardware components. The developed LWIP prototype is made to
work with commercial UE (Nexus 5). The LWA prototype requires modification at the UE
protocol stack, hence it is realized using OAI-UE. The developed prototypes are coupled
with the legacy multipath protocol such as MPTCP to investigate the coupling benefits.
Actions (login required)
 |
View Item |
Altmetric
Altmetric