SDN-Enabled Integrated Space-Air-Ground

Networks: Towards a Convergence

Nariman Torkzaban*, Anousheh Gholami*, John S. Baras*, Chrysa Papagianni**
*Institute for Systems Research,
Dept. of Electrical & Computer Engineering
University of Maryland College Park, USA
**University of Amsterdam
narimant@umd.edu, anousheh@umd.edu, baras@umd.edu, c.papagianni@uva.nl

3rd Workshop on 5G Technologies for First Responder

and Tactical Networks 2020- Research Track

1
 Outline
● 5G promises and use-cases
Supporting Technologies/Components:
○ Satellites & 5G
○ Aerial Platforms
○ SDN/NFV

● SDN-Enabled Space-Air-Ground Integrated Networks

Integration Challenges:
○ (Trust-Aware) Service Function Chain Embedding and Resource Allocation (SFC-RA)
○ Joint UAV Placement and Traffic Routing in Multi-Hop UAV Relaying Systems (JUPR)
○ Joint Satellite Gateway Placement and Routing for Integrated Satellite-Terrestrial Networks
(JSGPR)

2
 5G Use-Cases

[Source: NGMN]
[Source: ITU-R]

[Source: 5G PPP]

[Source: 3GPP]

3
 5G Vision
End-to-end ecosystem to enable a fully mobile and connected society Value creation towards
customers and partners, with existing and emerging use cases delivered with consistent
experience

5G Use Cases 5G Business

5G Value Creation Models

[Source:: A Survey of
5G Network:
Architecture and
Emerging Technologies,
IEEE Access, 2015]

Network of networks, i.e., a heterogeneous system comprising a variety of air interfaces, protocols,
frequency bands, access node classes, and network types

4
Satellite and 5G

5
 Satellite and 5G (Contd..)

[Source: G. Giambene, S. Kota and P. Pillai, "Satellite-5G Integration: A Network [Source: SES, The Role of Satellite in 5G, Satellite Connectivity
Perspective," in IEEE Network, vol. 32, no. 5, pp. 25-31, September/October 2018 ] Workshop.Nadi, Fiji, 24 April 2017]

6
 Satellite and 5G (Contd..)

[Source: Key elements for integration of satellite systems into Next Generation Access Technologies, ITU Report, Jul. 2019]

7
 Aerial Platforms
● UAV-aided wireless communications can fall into three representative categories of use cases:
(i) UAV-aided ubiquitous coverage (ii) UAV-aided information dissemination and data collection and
(iii) UAV-aided relaying.

● Main advantages: ● Main challenges:

- Fast deployment - Minimization of the deployment cost (the number of
- Cost and power efficiency UAVs, unsupported traffic, energy consumption, etc.)
- On-demand mobility - UAV placement and trajectory design
- Line of Sight (LoS)-dominant links - Both A2G and A2A connectivity, E2E traffic
delivery (joint facility placement and traffic routing)
- Spectrum utilization
- Interference management

8
 Network Slicing
5G network slices implemented on the same infrastructure

[Source: NGMN]

9
Network Slicing (Contd..)

[Source: TELCO 2.0. STL Partners- HP]

10
Network Function Virtualization
Examples of VNFs:
● Switching: BNG, CG-NAT, routers.
● Mobile network nodes: HLR/HSS, MME, SGSN, GGSN/PDN-
GW, RNC.
● Home routers and set top boxes.
● Traffic analysis: DPI.
● Signaling: SBCs, IMS.
● Network-wide functions: AAA servers, policy control.
● Application-level optimization: CDNs, Load Balancers.
● Security functions: Firewalls, intrusion detection systems.

11
E2E Network Slicing – SFC Allocation

12
Network Function Virtualization (Contd..)
● Network Service: An e2e functionality provided by an
operator, delivered using one or more (Virtual)
Network Functions.
● Service Function Chain or VNF-FG: Ordered set of
VNFs and ordering constraints that must be applied for
”steering” of traffic through them, to create a
composite service .
● Network Forwarding Path: indicates a traffic flow on
the network service.
● VNF: Function responsible for specific treatment of
data flows that can be implemented as a virtual
element or embedded in a physical appliance.
● NFV Infrastructure (NFVI): Hardware and software
required to deploy, manage and execute VNFs
including computation, networking and storage

13
Software Defined Networking (SDN)

14
 Space-Air-Ground Integrated Network
The space-air-ground integrated network (SAGIN) offers potential benefits which are not
possible otherwise, including global coverage, low latency and high reliability.

Main challenges of SAGIN include:

● Optimized integration tailored to specific use-cases and QoS requirements

● Optimal placement of different facilities such as satellite gateways, controllers, UAVs and allocation of
resources such as radio spectrum, power, etc.
● Mobility management and routing mechanism considering SAGIN’s heterogeneous architecture
● Traffic offloading and load balancing in the presence of several access technologies

[Source: J. Liu et al., “Space-Air-Ground Integrated Network: A Survey,” in IEEE Communications Surveys and Tutorials, vol. 20, no. 4, pp. 2714-2741, Fourthquarter 2018.]

15
Convergence: 5G, SDN, NFV, MEC, IoT,..

16
SDN-Enabled SAGIN

Towards achieving the 5G

key promises, it is essential to
utilize the capacity of all sorts
of communications networks
(terrestrial, space, aerial) and
supporting technologies
(SDN, NFV, etc.)
simultaneously, as opposed
to the traditional stand-alone
fashion.

17
Service Function Chain Embedding and
Resource Allocation (SFC-RA)

18
 Service Function Chain Embedding

[Source: Resource Allocation in NFV: A Comprehensive Survey, IEEE TNSM, 2016 ]

[Source: G. Wang, et al., "SFC-Based Service Provisioning for Reconfigurable Space-Air-Ground Integrated Networks," in IEEE Journal on Selected Areas in Communications, vol. 38,
no. 7, pp. 1478-1489, July 2020.]

19
SFC-Resource Allocation

20
SFC-Resource Allocation (Contd..)

21
SFC-Resource Allocation (Contd..)

22
 vEPC-NF Placement
Challenges
● Tight coupling between data and control plane elements
○ Delay budgets among cellular core elements
● Scale
○ Large number of UEs and DCs in cellular networks

Existing NF placement methods:

● Optimize the placement of data-plane functions

23
 Solution Approach

[Source: Dietrich, David., Papagianni, Chrysa. , Panagiotis, Papadimitriou., Baras, John S., "Network Function Placement on Virtualized Cellular Cores",COMSNETS 2017.]

24
 Trust-aware Service Chain Embedding (TASCE)
SFC Request Substrate SFC Embedding
GF GS Solution Trust weights:
• Indicating the trustworthiness of a system
• Hosts: A composite of multiple measures such as
location, interaction with other network entities,
level of hardening for the host, etc.
• VNFs: Required trust level

VNF placement: on which server each VNF will

be hosted and allocated physical
resources (computational power, memory, storage…)
Traffic steering : which communication paths
ep1, ep2 = end points u, w, x, y =
servers should be used to transfer data between each pair
i, j, k = VNFs
(ep1,k), (k,j), (j,i), (i,ep2) = traffic
v, z = routers of VNFs and bandwidth allocation ​
flows z = entry point
ti = trust weight tu = trust weight

25
 TASCE- Performance Evaluation
Service chains:
VNF forwarding graphs:
• Serial processing

• Traffic splitting

• Bifurcated path with single EP

[Source: N. Torkzaban, C. Papagianni and J. S. Baras, “Trust-Aware Service Chain Embedding,” SDS conference, Rome, Italy, 2019, pp. 242-247]
[Source: N. Torkzaban,, and J. Baras. "Trust-Aware Service Function Chain Embedding: A Path-Based Approach." arXiv preprint arXiv:2009.07343 (2020).]

26
 TASCE- Performance Evaluation (Contd..)

● The performance of the MILP is

superior in optimality.
● The LP has less time complexity.
● Resource utilization lags behind in the
LP approach due to sub-optimal
resource allocation.

● Approximately 10% less CPU and

BW revenue for LP

27
Joint UAV Placement and Traffic
Routing (JUPR) in Multi-Hop UAV
Relaying Systems

28
 UAVs as Flying Base Stations - Objective
● Multi-hop UAV relaying
systems in the next generation of
wireless networks results in the
coverage of more UEs and the
support of long-distance
communications, in addition to
the ability to handle overloaded
networks
● Use-cases: disaster relief, rural
areas, high demanded area

[Source: A. Gholami, et al. “Joint Mobility-Aware UAV Placement and Routing in Multi-Hop UAV Relaying Systems.” arXiv preprint arXiv:2009.14446 (2020).]
[Source: A.Gholami, et al. “Drone-assisted communications for remote areas and disaster relief”. arXiv preprint arXiv:1909.02150 (2019)]

29
Loon delivered emergency connectivity after
Hurricane Maria in Puerto Rico - 2017
30
 JUPR- Numerical Results
● Evaluation of the proposed solutions for a network of disconnected ground clusters with RPGM mobility model (a
good model for the movement of rescue teams during disaster relief):

UAV placement example: (a) proposed optimal (DM-MILP) (b) approximation (DM-LP) (c) Number of UAVs: Optimal vs
connectivity-based (DM-Conn) Approximation

● Only 67% of traffic demand is supported in DM-Conn (connectivity-based solution

that ignores the traffic and topology dynamics
● The proposed adaptive solution outperforms a static UAV placement method
(approximately 20% more supported traffic) while balancing the time complexity
and the optimality (5-7% less supported traffic compared to the optimal MILP)
Average percentage of the supported traffic

31
Joint Satellite Gateway Placement and
Routing for Integrated Satellite-
Terrestrial Networks (JSGPR)

32
 (JSGPR) Introduction
Joint Satellite Gateway Placement and Routing in an Integrated Satellite-Terrestrial Network
● Satellite Gateways as Relays to the Space Layer

● Capacitated Facility Placement + Multi-Commodity Flow Allocation

○ Usually addressed as two independent problems in a sequential manner
■ Neither very realistic, nor effective
■ Suboptimal Facility Placement may impose (huge) extra cost on flow allocation

○ The two problems share significant inter-relations

■ Makes it reasonable to approach them simultaneously
■ We propose a joint formulation for minimizing the overall cost of gateway placement and flow allocation

33
 JSGPR Introduction (Contd..)

● Hybrid (Integrated) Satellite-Terrestrial

Network (ISTN)

● Terrestrial Layer: Ground Network

consisting of 5GC, gNBs, end-users.

● Space Layer: Multi-Layer Satellite

Network (LEO, MEO, GEO) with
Global Coverage

34
 JSGPR- Performance Evaluation
✔ Exp. A : Benchmarking the MILP-based vs the LP-Based method (JSGPR MILP vs JSGPR LP)
✔ Exp. B : Evaluate the Impact of the Load Balancing technique (JSGPR-LB)

Multiple Real Network Topologies extracted from the publicly

available Internet Topology Zoo:

Metrics
• Solver Runtime
• Average Delay
• Total Cost
• Gateway Load

[Source: Knight, Simon, et al. "The internet topology zoo." IEEE Journal on Selected Areas in Communications 29.9 (2011): 1765-1775.]

35
Numerical Results for Gateway Deployment

36
 Joint Gateway Placement and Routing Results
● Exponential increase in the MILP solver
time with the size of the network.
● The LP-based approach results in additional
deployment costs within the range of at
most 13% of the optimal placement. The
gap is decreasing as the network size gets
larger.
● In Agis and Digex the suboptimal Average Solver Runtime (seconds) Normalized Average Total Cost

placement leads to the placement of

additional gateways; thus nodes end up
closer to the gateways and experience
lower expected delay.
● The higher the tolerable delay is, the lower
the total cost is. For instance, upgrading the
service from 2ms to 5ms in Agis will
almost double the cost.
Average Network Delay
Normalized Average Total Cost with varying delay bound

37
 Thank you!

Questions?
John S. Baras
Email: baras@umd.edu
Web page: https://johnbaras.com/

38