The implementation choices of very high data rate ad hoc networks for short and medium ranges, covering the physical, medium access control (MAC) and network layer are broad. Several proposals exist which use crosslayering techniques for the MAC and physical layers, but so far few proposed deeper cross-layering covering the three layers, and above. However, some optimizations, essential for the network and application layer performance, can only be achieved through this deeper integration.
Nodes must self-organize to control network access and reduce interference. Access can be organized into structured PANs (Personal Area Networks) composed of stable groups (scatternet), each one with one master node (e.g. 802.15.3); or it can be unstructured and decentralized (e.g. DCF (Distributed Coordination Function) of 802.11) for ad hoc medium access. Although DCF’s theoretical peak rate is high, the effective throughput available per user is much lower due to MAC inefficiency on sharing the channel and on carrying broadcast traffic. Nowadays, there is a strong interest in UWB (Ultra-Wide Band) transmission schemes for ad hoc networks. The UWB signals have huge bandwidths and small power spectral densities, even below the channel noise levels. Therefore, an UWB-based system can share the spectrum with several “narrow-band” systems with minimal performance degradation for them. Moreover, since the UWB schemes use spread spectrum techniques allowing very high processing gains (even for services with moderate or high data rates), they provide strong robustness against interferences, such as those associated to the multiple “narrow-band” signals sharing its bandwidth.
The impulse radio techniques are the most popular candidate for UWB transmission, typically employing selected pulse modulation schemes combined with TH-MA (Time-Hopping Multiple Access). This is especially due to the reduced implementation complexity of impulse radio, when compared with continuous-wave UWB options. However, since the spectral efficiencies achievable with impulse radio techniques are typically low, there is an increased interest on UWB systems employing continuous-wave techniques such as OFDM (Orthogonal Frequency Division Multiplexing), DS-CDMA (Direct Sequence Code Division Multiple Access) and MCCDMA (MultiCarrier CDMA).
This work considers transmission techniques for UWB-based ad hoc networks. Both impulse radio, combined with TH-MA, and continuous-wave schemes, namely OFDM, DS-CDMA and MC-CDMA schemes, will be considered for the UWB radio transmission. Appropriate signal processing schemes will be developed and evaluated, for both the transmitter and the receiver, so as to improve the range/bitrate tradeoffs.
Improved receivers, with multipath and multiaccess interference cancellation, will be developed. The use of multiple-antenna systems to improve the performances and/or to increase the capacity will be considered. The synchronization and channel estimation requirements will be studied, as well as appropriate estimation methods, namely employing iterative detection/estimation procedures will be developed.
For both transmission techniques (impulse radio and continuous-wave) we will consider DS (Direct Sequence) spreading and CS (Code Spread) schemes, namely employing TCH codes (Tomlinson Cercas Hughes).
Since the system should be able to share the spectrum with present narrow-band systems, there is especial interest in the evaluation of the mutual interference levels, as well as the development of techniques to minimize these interferences. To reduce the interference levels, we will consider appropriate pulse/spectral shaping techniques, as well as interference cancellation schemes.
Medium access control was traditionally implemented independently of the physical layer. Contention based models rely on CSMA/CA (Carrier Sense Multiple Access/Collision Avoidance) access mode, enhanced with RTS (Request to Send) / CTS (Clear to Send) mechanism for reducing the hidden node collision problem for large packets. To improve throughput and quality of service, priority mechanism and frame aggregation mechanism were introduced. However, it is not possible to offer QoS guarantees without introducing effective reservation mechanisms for broadcast and point-to-point communication. This can be achieved by coordinating access on a self-organized network, or by introducing cross-layering reservation mechanisms possibly supported by the physical layer (e.g. codes, bandwidths, time slots). From the network layer point of view, the final performance is also related to the relative performance of the broadcast traffic, since most of the basic routing and service discovery services use it. Broadcast traffic is more sensible to collisions since it is not acknowledged at the MAC layer. Its performance can be improved using cross-layering approaches. In this project we intend to optimize the throughput at the network layer, through an evaluation of the study of relevant implementation choices and crosslayer implementations.