Aerospace and Electronic Systems Magazine May 2018 - 46

Enabling RF Data Analytics Services and Applications via Cloudification
is composed of high-performance programmable processors and
real-time virtualization technology [11]. The connection between
the RF client and the cloud is not just a high-bandwidth connection. The interface must also support different types of RF clients,
with different capabilities, operating over different bandwidths. To
enable the connection of a myriad of different types of RF clients,
the interface must be open.
Cloud-based techniques called C-RAN have recently emerged
[8]-[11]. C-RAN technology allows the antennas, together with
the RF front-end circuitry, called remote radio head (RRH), to be
positioned so that coverage is optimum, while the digital subsystem can be located further away in a cloud. Software is deployed
in the cloud, and the LTE cellular standard is provided as a service
(aaS), with the RRH being the client. This architecture can lower
both capital expenditure and operational expenditure to deploy
current long-term evolution (LTE) cellular networks. Note that CRAN is a private cloud architecture, supporting only one service
LTE, offered to the RRH.
The explosive growth in wireless services has made the wireless infrastructure increasingly diverse. Various wireless technologies have different understandings of the radio spectrum. C-RAN
is entirely inappropriate in such a heterogeneous environment. Indeed, C-RAN is not even intended to offer services to the public.
Therefore, we advance RF cloud techniques that are public and
therefore based on an open interface. An open interface will allow any type of RF client to connect to any cloud. Previously, we
developed an interface [12] that contains metadata represented by
a single parameter, such as RF center frequency, bandwidth, and
location. To better support RF cloudification, we extend the interface [12] to include higher-level information and to associate this
higher-layer information with the already paired data and metadata. With this extension, the interface allows extension packets
containing arbitrarily defined metadata, which can include entire
files with ontology or XML descriptions. Data packets, metadata packets, and extension packets of arbitrarily defined data and
metadata can be exchanged over this interface, allowing applications anywhere on the Internet to use this cloud architecture.
Although in this article we refer to "the cloud," it is more generally a "cloud of clouds" or Intercloud [13]. Recently, it has been
noted that clouds of clouds have interoperability problems and
require appropriate interfaces to solve these problems [12], [13],
[14]. Unlike the interface used in C-RAN, as described in [12], the
open interface enables the movement of RF data from one cloud
to another and completely avoids these interoperability problems.
The interface is completely independent of the specific technique
to implement the interface. For example, optical fiber can be used
to implement the interface [11]. It is interesting that wireless fronthaul is also possible, referred to as radio over radio. There are
deployment scenarios, where such wireless fronthaul is more convenient and cheaper.
However, there is one significant bottleneck: the interface must
have high bandwidth to handle the very high data rate that is required. Some cloud services, such as LTE, require IQ data at a
very high rate. Recently, compression of IQ data has been studied
[11], [15] as a way to reduce the data rate required by C-RAN
over this interface. Spectrum monitoring, the service considered
46

further in this article, requires a small fraction of all spectrum data
to be sent over the interface, makeing the data rate much lower.
Note that the metadata packets, although very important, represent
a small overhead, because the parameters and the descriptions they
carry change at a much lower rate, especially compared with the
IQ samples.

RF CLOUD SERVICES AND RF DATA ANALYTICS
The RF cloud architecture described here allows a number of novel
services to be offered. Depending on the needs of the service users,
similar to cloud computing, these services range from infrastructure aaS (IaaS) and platform aaS (PaaS) to software aaS (SaaS).
However, the proposed RF cloud architecture allows, and even
requires, novel types of business relationships that do not exist in
models such as C-RAN and traditional cloud computing. The first
difference is that RF clients can also offer services, such as Iaas,
PaaS, and even some SaaS. The services offered by the clients are
typically different from the services offered by the cloud, but some
overlap is possible. The data rate at the interface depends on the
separation of services. It is possible that RF clients do some or all
of the signal processing. This substantially lowers the data rate at
the interface but decreases the flexibility and the opportunities for
cloud services.
Furthermore, some services such as spectrum monitoring are
composite services in the sense that they require the collaboration
of RF clients and clouds. The open architecture allows the RF client and a service user to be two different entities. A service provider can use a cloud provider and one or more RF clients and can
create novel and composite services. Similar to clouds, RF clients
can also be viewed as built from an abstract pool of resources,
i.e., there is radio virtualization. Therefore, the advanced open
architecture changes also the cloud economics, i.e., the costs and
benefits of RF clouds. The most demanding services both in terms
of data rate of the interface and processing are those that require
modulation or demodulation of the RF signal. As an alternative to
C-RAN RAN, aaS can be offered as a public service, on the basis
of the described open interface. This can be disruptive to C-RAN
and is an appropriate topic for future research.
In addition, cloudification according to the proposed architecture is not just about adopting the pay-as-you-go business model
that is fundamental to C-RAN and cloud computing in general.
This RF cloud architecture allows providing not just technology
aaS (whether hardware, platform, or software), but also the revenue model can also be based on data or outcomes aaS. Examples
of data aaS include RF spectrum desktop aaS, and eventually, as
the interface data rate allows, baseband IQ aaS. Outcomes aaS can
include spectrum usage information or abnormal spectral event
notification aaS.
We have developed a test cloud-based platform on the basis
of the general principles of the architecture described previously,
allowing different types of RF clients to be connected. In our prototype, shown in Figure 1, we have used different RF clients with
various capabilities. The RF clients are based on software-defined
radios (SDRs) and are connected via an optical backbone (highspeed Internet connection) to the cloud. Through open interfaces,

IEEE A&E SYSTEMS MAGAZINE

MAY - JUNE 2018

Table of Contents for the Digital Edition of Aerospace and Electronic Systems Magazine May 2018