This website is preserved for historical and scholarly reference and is no longer actively maintained.
last update: 9/2/2022 TBD: need to
update/2020
Low Latency Networking - Literature survey and
modeling assumption highlights
Topic 1 Propagation and fading
models for ns3
Wireless signals are subject to multiple forms of impairment as the
RF energy propagates from a transmitter to a receiver.
- Propagation loss - In the best case scenario, the
path between two stationary stations, one the sender (call it
Tx) and one the receiver (call it Rx) form a path that
is within line of site, exists in unobstructed
free space eliminating multipath interference, and is not
subject from interference by any other Tx stations.
In this (unrealistic) environment, a wireless channel can be
modeled by deterministic log distance models - refer to
the VANET-NS3
paper or the ns3 documentation on propagation
models for details.
- The ns3 tutorial third.cc program builds a wireless network
model including these lines:
- The friis propagation model is a simplified version of
the log distance model propagation models that do not require
measurements for calibration. It is based on the physics
of RF propagation in free space.
- In addition to long term fading, short term fading can
be caused if multiple echoes of the transmission are
received.
- The simplest model is two-ray reflection that considers the
direct signal along with the effects of a secondary signals
caused by ground reflection.
- Two models that are more commonly used in wireless studies
are Rayleigh Fading or the Nakagami propagation model.
As the former is actually a form of Nakagami, we focus on the
latter model.
- The Nakagami propagation model addresses fading due to
multipath. An additional loss model is required to
accurately model large scale propagation
effects. The Nakagami model provides a parameter
(m) that can represent three distances each of which
contribute to fading impairment on the energy that arrives
after large scale propagation is accounted
for. Refer to the papers VANET-NS3
and VANET-FADING
for examples of appropriate settings.
- Another form of impairment that needs to be considered is
interference from other transmitting stations (or from the same
station if equipped with multiple radios). We
will focus primarily on the effects of transmissions by other
stations using the same (or close to) the frequency used
by a Tx station under observation. The main effect
is that interference caused by nearby competing stations
increases the noise floor of the channel. The typical measure of
the effects is the signal-to-interference-plus-noise ratio
(SINR). Theoretic models exist that provide upper
bounds on channel capacity given a particular SINR,
expected Rx power and modulation and coding. The YANS
paper presents a clear explanation of the basic 802.11 model
used in both ns2 and ns3.
- A common technique that is used in wireless research is to
utilize stochastic techniques that model the success or failure
of an Rx capture process outcome. This can be used in
conjunction with 'realistic' models such as Nakagami that try to
realistically RF energy impairment in targeted environments
(e.g., rural or urban). Or, the network
model can be constructed such that all loss is caused by the
artificial loss process model. Refer to this dissertation
for a straight forward discussion of applying stochastic
loss models to a wireless channel (they cover both a Bernoulli
loss model and a Gilbert Elliot model).
- For our initial studies:
- TODO: A brief comparison of ns3 results with either
a VANET or MANET prototypes. Our goal is to
identify perhaps two settings that represent a rural (or open
highway) and an urban settings. We need
a better understanding of the RF capture model NS3 uses
- I am guessing there are l dependencies between choices
of prop and loss models.
- TODO: We are likely to utilize NS3's artificial loss
models...we need a baseline script that we can validate.
- TODO: List the other default ns3 settings (Tx
power, MCS, 802.11a or 802.11n (and MIMO
assumptions), ....:
Topic 2: High level summary of the
DOD's direction for future battlefields
Future DoD systems will likely reflect Information Centric
Networking concepts. Much of the literature assumes this implies
forms of NDN. There seems to be a desire to use SDN.
Much of the original enthusiasm for SDN in academic communities has
diminished (who wants to spend a year of a student's time
dealing with lousy SDN open source code). In the
real-world, SDN was of interest only because it potentially
provided a standard configuration and monitoring interface - however
that interest seems to have peaked. The DoD community is
likely more interested in future battlefields that have a separate
data and control plane - which can be met through spectrum
management or standard wireless 'slicing' methods.
IOBT:
The Internet of Battle Things (IOBT) reflects the possible adoption
by the DoD of evolving Internet concepts such as the Internet of
Things, Edge computing to support a battlefield that in one
dimension is a complex system of machines and sensing devices that
integrate forming a cohesive distributed computing environment.
SDN-DTN
NDN-Tactical
CurrentStateOfMulticastRouting
Topic 3: High level summary of ICN
concepts
ONTOLOGIES
Military Ontologies for Information Dissemination at the
Tactical Edge and other similar papers describe the DoD's equivalent
direction to the Internet's quest for The Semantic Web [ ].
Core to ICN is the choice of data and how it is represented.
The eXtensible Markup Language (XML) and the Resource Description
Framework (RDF) provided techniques to tag data and to identify more
complex associations of particular data along with an underlying
ontology that ties identifiers into a common language are the basic
components required.
Survey-ICN
, RECENT-ICN:
Various ideas for a data oriented Internet began to appear in
the literature beginning in 1999. In 2007 the NSF began
funding research specifically focusing on ICN. One approach
that was widely studied was Named Data Networking (NDN) [NDN,
NDN-Brief-Intro]. NDN is a request driven architecture
where receivers express interest in particular data to NDN
routers. The interest message contains information as to the
source (producer) of the data allowing NDN routers to forward
Internets to producer nodes. NDN routers note who is interested in
particular data storing enough information and collectively
sufficient information within the Internet so that when the producer
creates a data message that matches a consumer's interest, the
message is forwarded through the Internet through the path of NDN
routers with each router ensuring the data is distributed to all
consumers that have expressed interest. The technique is
equivalent to the Internet's current unicast and multicast
communications methods. NDN is clearly Internet-oriented and
is appropriate for large messages (or perhaps streams of data) that
are to persist within the Internet for long time periods. Caching of
data at NDN routers promote scalability.
PUBSUB
Closely tied to ICN is the distributed communications method
referred to as publish-subscribe (or pubsub). As the use of
synchronous-oriented RPC was found to have scalability issues in
distributed systems, the concept of asynchronous message oriented
middleware (MOM) was established and widely used. Fundamental to MOM
systems is that messages are independent and contain all the
necessary information to allow an appropriate system to deliver the
message to one or more receivers. MOM incorporates message
queues with possible options for reliability and quality of
service.
A specific form of MOM was developed by IBM in 1999 known as MQ
Telemetry Transport (MQTT). This was originally developed to
create a lightweight bandwidth-efficient messaging protocol that was
data agnostic with support for optional support for reliability and
quality of service. Typically used in an asynchronous manner,
clients that serve as data sources publish data to a centralized
broker. Clients that serve as receivers of data subscribe to
data that they are interested in receiving to a broker.
Messages contain any number of topics (and subtopics) supporting the
dissemination of very specific data or, with support of
wildcards, the dissemination of aggregated topics to
interested subscribers.
A
DHT-based Infrastructure for Content-based Publish/subscribe
Services
Publish/Subscribe
Versus Content-based Approach for information Dissemination
Towards
Efficient Publish-subscribe Middleware in the IoT with IPV6
DM-MQTT
: And Efficient MQTT based on SDN Multicast for Massive IoT
Communications
An
SDN-MQTT Based Communication System for Battlefield UAV Swarms
Recent work by the research community has explored scalable pub-sub
typically by leveraging geo information in the data
Location-Aware
Pub/Sub System: When Continuous Moving Queries Meet Dynamic Event
Streams
GeoBroker:
Leveraging Geo-Contexts for IoT Data Distribution
Summary: Much of the ICN work is simply applying NDN concepts.
NDN and pub/sub was invented for the Internet. NDN in
particular is best suited for large data messages that persist for
long time periods. For example, NDN could potentially
work well for Internet streaming providers. Pub/Sub is
becoming the dominant method for disseminating IoT data.
However, NDN and pub/sub will likely not achieve the original goals
until the Internet figures out a way to promote the sharing and
reuse of data. Recent work related to location aware pub/sub
is somewhat relevant as it addresses how to deal with large streams
of data disseminated to mobile consumers.
In a wireless context, there are There are a few papers
that address the core issues of interest to us - the blending
of pub/sub with the challenges of highly mobile systems.
Location aware pub/sub can be re-looked at with the assumption that
both the sender and receiver are mobile. It seems that
application systems that require highly functional, highly adaptive
wireless systems continue to be limited by wireless issues.
Our direction for embellishing current adhoc wireless technologies
with 1. Separation control and data planes to all nodes;
2. Co-adaptation by application, middleware, infrastructure as the
system performance evolves; 3. Blending pub/sub with traditional
adhoc wireless to adapt the system on demand based on the data
Topic 4: High level summary of
modeling or studies of UAV/drone systems
Divide and conquer....
- Alarming:
- High level surveys
- Mobility Models
- UAV control (motion, flight plan, model-based)
- Cooperative UAVs
- UAV/Swarm communications, routing
- UAV simulators/emulator systems
Summary: Research related to UAVs and Drones has been
underway for 15 or more years. Recent work includes
collaborative drones performing tasks, frameworks that provide
platforms to simulate, emulate, or prototype experimental drone
controllers or drone distributed applications. As far as
networking, there is no clear direction. Many systems
utilize traditional adhoc systems using combinations of older and
newer routing protocols.
There are several papers that talk about the use of ICN or pub/sub
in drone swarms. This is likely an area that will
attract more interest.
************************************************************
- Overview and links to papers
- Topical Areas