Implementation of a cognitive radio modem

Graduate students: Peiman Amini, Ehsan Azarnasab, Salam Akoum, Xuehong Mao, Harsha Rao.

Construction of a cognitive network for first responders presents many difficult challenges, the most obvious of which is how to fulfill the “awareness” requirement. Each node must be able to sense the channel for primary users, i.e. to identify the presence of licensed communications over the portions of spectrum, and share this information with the other nodes to allow the cognitive nodes to communicate
reliably while avoiding the legacy devices. The method used for sensing should feature a high spectral dynamic range to enable the detection of the low power users. The cognitive radio transceiver should also feature a timing and carrier synchronization method that is efficient and cost effective in terms of resources available.

We investigate different software radio algorithms to come up with an efficient implementation of a cognitive radio transceiver to be used in disaster scenarios. This on-going research is part of the Software Radio Challenge hosted by the SDR Forum. Read more

 

MIMO/CDMA Detection techniques

Graduate students: Xuehong Mao, Salam Akoum, Andy Laraway, Peiman Amini.

We have developed/ are currently developing a number of Bayesian detection methods that are applicable to both synchronous code division multiple access (CDMA) and multiple-input multiple-output (MIMO) communication systems. Markov chain Monte Carlo (MCMC) simulation techniques are used to obtain Bayesian estimates (soft information) of the transmitted bits. We have demonstrated that such detectors perform very close to channel capacity. Comparisons with sphere decoding and previous MCMC based method have shown that our detectors are superior.

Our ongoing research includes the followings:

  • Hardware architecture for implementation of the MCMC detector.
  • Code design for near capacity performance.
  • Performance study of MCMC detector in high SNR regime.
  • MCMC detector performance in MIMO-OFDM.

 

Filter-Bank Based Multitone For Cognitve Radios.

Graduate students: Peiman Amini.

We study methods of using filter banks for multicarrier communication in a Cognitive Radio (CR) setup. Two solutions are investigated. The first solution uses subcarrier bands that are non-overlapping. The method is referred to as filter multitone (FMT) and was originally developed for bi-directional transmission of data over digital subscriber lines (DSL).  From the bandwidth efficiency point of view, this solution may not be attractive because of guard/transition bands between adjacent subcarriers. However, it offers advantages from the simplicity point of view.  The second solution uses cosine modulated filter bank. We call this cosine modulated multitone (CMT). This method also has roots in DSL, and has recently been revisited and applied to wireless applications as well.  This solution offers the advantages of bandwidth efficiency and blind equalization capability. When multiple adjacent bands are used to carry the data of one user, overlapped adjacent bands can be separated perfectly thanks to perfect reconstruction property of CMFB. We also develop random medium access control (MAC) protocols for the proposed CRs.

 

 

Synchronization and Modeling of Mobile OFDM.

Graduate Student: Scott Talbot

Mobility presents several challenges for OFDM modulation.  One of the greatest problems is the interference that is introduced among the different OFDM subcarriers as a result of mobility-induced time-variations in the channel.  Since the communication channel is one of the largest sources of communication performance degradation, a lot of previous research has focused on channel characterization and modeling.  Our work has focused on characterizing OFDM ICI for common channel models that have been developed due to this research and that go beyond the standard Rayleigh-fading Clarke's (Jakes) Doppler spectrum model.  Moreover, system mobility is also a source of carrier offsets in OFDM due to Doppler, and additional research has focused on synchronization algorithms for these types of carrier offsets and on developing appropriate models for Doppler carrier offsets since they can be time-varying and are more complicated than carrier offsets due to oscillator inaccuracies.

 

Wireless CDMA Ad-Hoc Networks.

Graduate Students: Ehsan Azarnasab, Roland Kempter

CDMA detection requires the knowledge of the user-specific spreading  sequences. Thus, if additional paging is impossible or unwanted such  as in (mobile) CDMA Ad-hoc networks, typically Spread ALOHA is believed to be the only means for data transmission. However, Spread ALOHA is collision and interference limited and multiuser detection cannot be used to improve system performance. As a solution, we present a novel fully asynchronous packet transmission system termed Random Packet CDMA. In Random Packet CDMA (or RP-CDMA), every packet consists of a header frame which is spread with a common sequence as well as a data portion with random spreading. Information which allows to recover the random sequence is contained in the header portion of the packet which is also used for packet specific timing recovery. As a result, RP-CDMA is no longer collision limited and multiuser detection can be used to improve system performance. In the next step, we introduce a novel CDMA detection scheme called partitioned spreading based on iterative decoding. In contrast to other CDMA detection schemes, partitioned spreading detection is near-far resistant. We compare partitioned spreading to the matched filter receiver, the decorrelator, the MMSE filter and a successive cancellation detector in randomly generated networks as well as in an Ad-hoc scenario based on a snapshot of the FLUX Mobile Robot testbed. We show that as the load increases, only partitioned spreading detection is able to maintain original network coverage and prevent network fragmentation. As a result, we motivate that in Ad-hoc networks, advanced joint detection might be a necessity -- not only to achieve higher transmission rates but merely to maintain connectivity.

 

Stereo Acoustic Echo Cancellation

Graduate students: Harsha Rao.

The focus of this research will be to explore the use of adaptive algorithms to implement stereo
acoustic echo cancellation. The system will also aim to achieve sound spatialization using head-related
transfer functions with acoustic crosstalk cancellation.