18-04-2012, 11:28 AM
Opportunities and Challenges for SDR in Next Generation TETRA Systems
Introduction
Software defined radio (SDR) technologies have been long promoted for use in the
professional mobile radio (PMR) application domain, as typified by the TETRA
standard. The need to retain and enhance compatibility with a proliferation of legacy
systems provided the incentive, and the typically very narrowband nature of these
communications schemes made SDR appear feasible. In practice the evolution of
these standards has moved faster than the development of viable SDR
implementations. In recent years, use of PMR radio systems has expanded beyond
voice to include data. There is now increasing demand for high speed data access to
support applications such as the transmission of highresolution
images (eg maps for
emergency services), and video clips. This has resulted in an evolution in the TETRA
standard, from the original voiceonly
service, to an enhanced data version (TEDS)
and finally to a proposed new approach which implements a WiMAX subchannel
within the TETRA band. This paper will analyse the effect that this evolution in the
standards will have on any potential SDR implementation.
TETRA Basestations Demand SDR
As in civilian public mobile radio, cost pressures and resource constraints on the
clientside
unit are high and this will result in limited opportunities for software
defined radio architectures. However at the basestation the situation is different.
Ideally a basestation should support existing and legacy devices, be frequencyflexible
so that it can be deployed globally, capture the full band so that frequency
reuse in different cells can be supported with a single hardware platform, and be
software upgradeable to support new standards. These are features that can be
offered by reconfigurable radio platforms incorporating software radio. As the
original communication schemes are very narrowband (25 kHz channels) with
simplistic modulation schemes, software radio systems could be implemented with
minimal complexity. As early as 2001, commercial implementations were available
demonstrating the flexibility offered by a software radio approach and commercial
software based solution for basestations are predicted for deployment in 2009.
Opportunities of the New TETRA Evolutions
A new release of the TETRA standard has just recently been adopted, called the
TETRA Enhanced Data Service (TEDS) which provides backward compatibility but
allows subchannel
bandwidths of up to 150 kHz, a tuning range of operating
frequencies between 350 and 470 MHz, and a number of modulation schemes. TEDS
can offer data rates up to 384 kbps (or higher in special cases). Deploying TEDS has
been a slow process due to the lack of a unified spectrum allocation, and the concept
of a tuning range for devices was proposed. This approach is ideally suited to
softwaredefined
radio where a wideband radio front end can be used. The
alternative is a multiband
radio which may be acceptable at the handset but is not
economical or viable at the basestation. Software defined radios also offer the
possibility of dynamically adapting the subchannel
bandwidth (from 25 – 150 kHz)
and modulation scheme as needed to support the needs of the users. These two
options strongly support the use of SDR techniques.
Beyond TEDS, new broadband TETRA schemes are being considered, including the
deployment of a WiMAX stack within a TETRA band, utilizing one or more WiMAX
1.25 MHz channels within the 5 MHz allocation. The benefits of this approach is
increased spectral efficiency and the ability to offer individual users a peak data rate
of over 3 Mbps as needed. Such a scheme must remain within the frequency
constraints of TEDS and once again a tuning range is required. With a given radio
front end, assuming a full band (5 MHz) could be captured, an SDR implementation
would allow a seamless upgrade path for TEDS devices to any of the newly
suggested TETRA evolutions.
Hardware Challenges
The hardware challenges can be grouped into a number of categories: frequency
agility; receiver sensitivity; backward compatibility; channelisation; and interfacing
with the software engine. The tuning range at up to 120 MHz corresponds to
approximately 30% of the centre frequency which is challenging when trying to
design selective but wideband filters or achieving low noise oscillators. As a result of
these challenges, it is likely that TEDS and broadband Tetra will be initially limited to
the range 380430
MHz. Noise and receiver sensitivity are particular challenges for
any TETRA system. The concept is that TETRA networks are light overlays over
existing networks providing secure backup communications for low numbers of users
in large cells. For this reason receiver sensitivity is very high with tight specifications
on adjacent channel interferers and noise. Within a narrowband system (25 kHz)
this was achievable but noise and vulnerability to interference will increase with
wider channels. Software radio normally implies channelisation in the software
domain however this can place excessive requirements on the software, particularly
where there are many subchannels
of different sizes within a TETRA band. If
channelisation is to be done in software, then this requires the full 5 MHz band to be
captured and presented over the PC interface to the software engine. Quick
calculations suggest that this is at the limit of current generalpurposeprocessor
capabilities. One alternative is to use a dedicated processor or FPGA on a hardware
platform to channelise the data, easing the computational load. Finally the interface
between the radio boards and the software engine can also be a major limiting
factor. Existing techniques, such as USB2.0 or Ethernet, face difficulties in ensuring
sustained latencyfree
communications of sufficient bandwidth for the full 5 MHz
TETRA bandwidth.
Software Challenges
Approximate calculations based on the information from [4] indicate that
the signal processing for a single 1.25 MHz WiMAX channel could be handled by a
modern general purpose processor (GPP). For example one core of an Intel Core Duo
using a well optimized FFT routine can execute more than 1.5 million 128 point
FFTs/sec [5] (equivalent to 192 Msps of raw I and Q data).
Unfortunately TETRA and WiMAX use different channel spacings which makes
channelisation more complicated and more compute intensive. Furthermore, existing
TETRA and WiMAX stacks use different media access schedules so the radio front end
must run in full duplex mode. Taken together these factors suggest that the raw
sample rate for even a single 5MHz band would exceed the input/output capabilities
of USB and even FireWire 800. Identifying a suitable high speed interface is still a
matter of investigation. However, as argued above, it may be more technically
feasible to perform the channelisation before input to the GPP although this may
result in some loss of flexibility and reconfigurability.
Even with the capability to modulate and channelise the radio waveform,
constraining the latency in the input/output and higher level software processing, to
the limits required for timely signaling and control, remains a significant challenge.
Conclusions
The technical challenges facing SDR implementations for PMR/TETRA systems are
now beginning to approach the complexity of existing mobile telephony systems.
Existing low cost SDR platforms and pcbased
solutions would find it challenging to
support TEDS or the next generation of TETRA, software defined radio architectures
present a compelling case for use in TETRA and other PMR systems. As with normal
mobile communications, to achieve a viable costeffective
solution a new approach to
solving the architectural in the overall platform is required which optimally partitions
resources between the radio circuits, dedicated hardware processors, interface
systems and the software engine and processor.
Introduction
Software defined radio (SDR) technologies have been long promoted for use in the
professional mobile radio (PMR) application domain, as typified by the TETRA
standard. The need to retain and enhance compatibility with a proliferation of legacy
systems provided the incentive, and the typically very narrowband nature of these
communications schemes made SDR appear feasible. In practice the evolution of
these standards has moved faster than the development of viable SDR
implementations. In recent years, use of PMR radio systems has expanded beyond
voice to include data. There is now increasing demand for high speed data access to
support applications such as the transmission of highresolution
images (eg maps for
emergency services), and video clips. This has resulted in an evolution in the TETRA
standard, from the original voiceonly
service, to an enhanced data version (TEDS)
and finally to a proposed new approach which implements a WiMAX subchannel
within the TETRA band. This paper will analyse the effect that this evolution in the
standards will have on any potential SDR implementation.
TETRA Basestations Demand SDR
As in civilian public mobile radio, cost pressures and resource constraints on the
clientside
unit are high and this will result in limited opportunities for software
defined radio architectures. However at the basestation the situation is different.
Ideally a basestation should support existing and legacy devices, be frequencyflexible
so that it can be deployed globally, capture the full band so that frequency
reuse in different cells can be supported with a single hardware platform, and be
software upgradeable to support new standards. These are features that can be
offered by reconfigurable radio platforms incorporating software radio. As the
original communication schemes are very narrowband (25 kHz channels) with
simplistic modulation schemes, software radio systems could be implemented with
minimal complexity. As early as 2001, commercial implementations were available
demonstrating the flexibility offered by a software radio approach and commercial
software based solution for basestations are predicted for deployment in 2009.
Opportunities of the New TETRA Evolutions
A new release of the TETRA standard has just recently been adopted, called the
TETRA Enhanced Data Service (TEDS) which provides backward compatibility but
allows subchannel
bandwidths of up to 150 kHz, a tuning range of operating
frequencies between 350 and 470 MHz, and a number of modulation schemes. TEDS
can offer data rates up to 384 kbps (or higher in special cases). Deploying TEDS has
been a slow process due to the lack of a unified spectrum allocation, and the concept
of a tuning range for devices was proposed. This approach is ideally suited to
softwaredefined
radio where a wideband radio front end can be used. The
alternative is a multiband
radio which may be acceptable at the handset but is not
economical or viable at the basestation. Software defined radios also offer the
possibility of dynamically adapting the subchannel
bandwidth (from 25 – 150 kHz)
and modulation scheme as needed to support the needs of the users. These two
options strongly support the use of SDR techniques.
Beyond TEDS, new broadband TETRA schemes are being considered, including the
deployment of a WiMAX stack within a TETRA band, utilizing one or more WiMAX
1.25 MHz channels within the 5 MHz allocation. The benefits of this approach is
increased spectral efficiency and the ability to offer individual users a peak data rate
of over 3 Mbps as needed. Such a scheme must remain within the frequency
constraints of TEDS and once again a tuning range is required. With a given radio
front end, assuming a full band (5 MHz) could be captured, an SDR implementation
would allow a seamless upgrade path for TEDS devices to any of the newly
suggested TETRA evolutions.
Hardware Challenges
The hardware challenges can be grouped into a number of categories: frequency
agility; receiver sensitivity; backward compatibility; channelisation; and interfacing
with the software engine. The tuning range at up to 120 MHz corresponds to
approximately 30% of the centre frequency which is challenging when trying to
design selective but wideband filters or achieving low noise oscillators. As a result of
these challenges, it is likely that TEDS and broadband Tetra will be initially limited to
the range 380430
MHz. Noise and receiver sensitivity are particular challenges for
any TETRA system. The concept is that TETRA networks are light overlays over
existing networks providing secure backup communications for low numbers of users
in large cells. For this reason receiver sensitivity is very high with tight specifications
on adjacent channel interferers and noise. Within a narrowband system (25 kHz)
this was achievable but noise and vulnerability to interference will increase with
wider channels. Software radio normally implies channelisation in the software
domain however this can place excessive requirements on the software, particularly
where there are many subchannels
of different sizes within a TETRA band. If
channelisation is to be done in software, then this requires the full 5 MHz band to be
captured and presented over the PC interface to the software engine. Quick
calculations suggest that this is at the limit of current generalpurposeprocessor
capabilities. One alternative is to use a dedicated processor or FPGA on a hardware
platform to channelise the data, easing the computational load. Finally the interface
between the radio boards and the software engine can also be a major limiting
factor. Existing techniques, such as USB2.0 or Ethernet, face difficulties in ensuring
sustained latencyfree
communications of sufficient bandwidth for the full 5 MHz
TETRA bandwidth.
Software Challenges
Approximate calculations based on the information from [4] indicate that
the signal processing for a single 1.25 MHz WiMAX channel could be handled by a
modern general purpose processor (GPP). For example one core of an Intel Core Duo
using a well optimized FFT routine can execute more than 1.5 million 128 point
FFTs/sec [5] (equivalent to 192 Msps of raw I and Q data).
Unfortunately TETRA and WiMAX use different channel spacings which makes
channelisation more complicated and more compute intensive. Furthermore, existing
TETRA and WiMAX stacks use different media access schedules so the radio front end
must run in full duplex mode. Taken together these factors suggest that the raw
sample rate for even a single 5MHz band would exceed the input/output capabilities
of USB and even FireWire 800. Identifying a suitable high speed interface is still a
matter of investigation. However, as argued above, it may be more technically
feasible to perform the channelisation before input to the GPP although this may
result in some loss of flexibility and reconfigurability.
Even with the capability to modulate and channelise the radio waveform,
constraining the latency in the input/output and higher level software processing, to
the limits required for timely signaling and control, remains a significant challenge.
Conclusions
The technical challenges facing SDR implementations for PMR/TETRA systems are
now beginning to approach the complexity of existing mobile telephony systems.
Existing low cost SDR platforms and pcbased
solutions would find it challenging to
support TEDS or the next generation of TETRA, software defined radio architectures
present a compelling case for use in TETRA and other PMR systems. As with normal
mobile communications, to achieve a viable costeffective
solution a new approach to
solving the architectural in the overall platform is required which optimally partitions
resources between the radio circuits, dedicated hardware processors, interface
systems and the software engine and processor.