uwicore. Ubiquitous Wireless Communications Research LaboratoryUwicore




One of the key targets of the m-Hop project is to improve the current state of the art by developing advance hardware testbeds that will provide unique and valuable insights into the operation, performance and configuration of multi-hop cellular systems based on mobile relays. In this context, the project has developed two unique research testbeds. While the first one is devoted to investigating in real-time the connectivity and end-to-end performance of MCN-MR systems under various operating conditions, the second testbed implements in a SDR platform the IEEE 802.11 standard to allow studies on advanced multi-hop communications and networking solutions.

Testbed 1: Multi-hop Connectivity in MCN-MR Systems

m-Hop has developed the first MCN-MR testbed developed to investigate the performance improvements that can be achieved through MCN networks using mobile relays through field tests, and the operating conditions under which such improvements can be achieved. One of the objectives of the testbed is then to investigate the performance benefits of MCN-MR over traditional cellular systems. As a result, the platform includes two cellular links with the tools necessary to monitor and evaluate their performance. One of these links is part of a MCN-MR connection, while the other one represents a conventional single-hop cellular link with which to compare the performance of MCN-MR. The platform also integrates 802.11-based ad-hoc relaying nodes. In addition, one of the ad-hoc relaying nodes acts as a bridge between cellular and ad-hoc technologies in the MCN-MR communications link. This node is capable of forwarding in real-time the transmitted data between the radio technologies without downgrading the overall network performance.

The cellular mobile station is a Nokia 6720c handset that supports the radio access technologies GSM/EDGE and UMTS/HSDPA. The Symbian-base terminal also incorporates the Nemo Handy application, which provides the terminal with a powerful radio monitoring capability. The processing of the logged measured data offers a valuable set of Key Performance Indicators (throughput, BLER, RSSI) that have been very valuable to analyze the cellular links QoS in the MCN-MR testbed.

The ad-hoc mobile nodes have currently been implemented using conventional laptops. The added wireless interfaces are in charge of the ad-hoc and multi-hop transmission and reception of packets. The chosen additional wireless interface is a wireless ExpressCard with Atheros chipset. This solution was chosen due its outdoor range, date rate, reliability and capacity to operate under ad-hoc mode using IEEE 802.11a/b/g/n. On the other hand, the laptop built-in wireless interface is in charge of capturing the transmitted and received packets by the added external wireless interface. This capturing capability allows continuously monitoring the performance of the ad-hoc 802.11 links. The MCN-MR testbed uses both Kismet and Wireshark software tools due to their distinctive features. Kismet is capable to time and geo-reference all the transmitted packets using the GPS data into a single log file. Wireshark offers powerful tools to organize and filter the captured network traffic. The hybrid nodes are those in charge of acting as a gateway between the cellular and 802.11 multi-hop ad-hoc networks. The node uses a Nokia 6720c terminal as modem to provide an HSDPA cellular link, and uses the wireless interfaces of the ad-hoc node to enable its 802.11 multi-hop connectivity.

The following figure represents an example of the deployment of the m-Hop connectivity testbed at the campus of the University Miguel Hernandez of Elche (UMH) in Spain to compare the performance of a conventional/direct cellular link (L2) against that achieved by the MCN-MR link (L1 mH1 mH2).



An example of the field test results comparing the performance of MCN-MR systems against that achieved using traditional cellular systems is shown in the figure below for outdoor environments at the UMH campus (additional tests for outdoor-indoor environments, among others, have also been conducted). The figure includes plots of the throughput measured over time. In the conducted field tests, the performance of a traditional cellular link with NLOS conditions with its serving BS is compared to that achieved with a MCN-MR link operating under LOS conditions through various hops. The field trials have been conducted downloading large-size files from a HTTP server located and managed at the Uwicore laboratory.



The single-hop cellular link (L2) experiences NLOS propagation conditions which results in the use of robust QPSK modulation and continuous link outage. On the other hand, the LOS propagation conditions experienced by the cellular link L1 significantly increases the throughput measured compared to that measured over the link L2. These improvements are at the origin of the higher throughput measured in L1, and the fact that the file was downloaded in just 90 seconds compared to the 500 seconds necessary through the single-hop cellular link L2. The throughput measured in each of the 802.11 ad-hoc wireless links is also shown in the figure. Although the 802.11 links are characterized by a higher theoretical throughput than the cellular links, the results illustrated in mH1 and mH2 links show similar throughput levels to those measured in L1. This is due to the real-time forwarding of the data by all the nodes participating in the MCN-MR link. Such real-time operation results in that the cellular link L1 represents the overall bottleneck of the MCN-MR connection. The measured average end-to-end throughput also highlights significant performance differences between the single-hop NLOS cellular link (229kbit/s), and the MCN-MR connection avoiding NLOS conditions through multiple LOS hops (816kbit/s).



Testbed 2: Fully programmable SDR IEEE 802.11 testbed

A second testbed has also been developed within the m-Hop project to conduct advanced studies on communications and networking protocols for multi-hop wireless communications. In particular, the project has implemented a fully programmable Software Defined Radio (SDR) implementation of the IEEE 802.11 MAC that can be used, configured and fully modified. The platform is built using USRP2 nodes developed by the Ettus Research, and illustrated in the photo below, and GNU radio.


USRP2 is a SDR platform that implements the front-end functionality, and the Analog to Digital (A/D) and D/A conversion on a FPGA. The physical layer processing is done on a PC where the USRP2 is plugged. USRP2 is connected to the PC through a Gigabit Ethernet interface that supports simultaneous I/O signals of 50MHz radio frequency bandwidth. USRP2 also provides fast and precise AD/DA converters, and a FPGA optimized for Digital Signal Processing (DSP) applications. It allows processing complex waveforms at higher sample rates, turning USRP2 into an appropriate platform for researching wireless ad-hoc functionalities. The USRP2 board incorporates a XCVR2450 dual band (2.4-2.5 GHz, 4.9-5.85 GHz) transceiver and a VER2450 antenna. GNU Radio is an open-source collection of signal processing blocks used for building SDR platforms. In fact, GNU Radio is the primary software platform supporting the PC drivers for USRP. The basic principle behind the use of GNU Radio is to build a Flow-Graph (FG) composed of various signal processing blocks that perform specific radio functionalities.

Different projects have been developing over the last years interesting SDR platforms over which to investigate advanced ad-hoc communications and networking techniques. However, to date, there was a lack of a generic 802.11 MAC protocol USRP2 implementation that can be used independently of the PHY layer. The mHOP project has then implemented a MAC layer that performs the contention service or Distributed Coordination Function (DCF), which is based on CSMA with Collision Avoidance (CA) technique, to regulate the access to the shared wireless channel. The implemented MAC also includes the carrier sense function to determine whether the channel is idle during a DCF Inter Frame Space period, and the back off process to avoid packet collisions when several nodes try to send a packet at the same time. In addition, the MAC implementation also integrates the RTS/CTS mechanism to solve the hidden and exposed terminal problems of wireless networks, and to reserve the channel through the virtual carrier sensing or Network Allocation Vector (NAV). The current implementation also includes the network discovery process of mesh networks that is characteristic of the IEEE 802.11s standard. This process is enabled through the periodic (with configurable intervals) broadcast exchange of beaconing messages among neighboring nodes. The figure below illustrates the finite-state machine of the implemented MAC layer; the top of the figure represents the MAC state transition upon receiving a frame from the PHY layer, and the bottom part of the figure illustrates the MAC layer contention processes required to transmit a frame over the wireless channel.

Different projects have been developing over the last years interesting SDR platforms over which to investigate advanced ad-hoc communications and networking techniques. However, to date, there was a lack of a generic 802.11 MAC protocol USRP2 implementation that can be used independently of the PHY layer. The m-Hop project has then implemented a MAC layer that performs the contention service or Distributed Coordination Function (DCF), which is based on CSMA with Collision Avoidance (CA) technique, to regulate the access to the shared wireless channel. The implemented MAC also includes the carrier sense function to determine whether the channel is idle during a DCF Inter Frame Space period, and the back off process to avoid packet collisions when several nodes try to send a packet at the same time. In addition, the MAC implementation also integrates the RTS/CTS mechanism to solve the hidden and exposed terminal problems of wireless networks, and to reserve the channel through the virtual carrier sensing or Network Allocation Vector (NAV). The current implementation also includes the network discovery process of mesh networks that is characteristic of the IEEE 802.11s standard. This process is enabled through the periodic (with configurable intervals) broadcast exchange of beaconing messages among neighboring nodes. The figure below illustrates the finite-state machine of the implemented MAC layer; the top of the figure represents the MAC state transition upon receiving a frame from the PHY layer, and the bottom part of the figure illustrates the MAC layer contention processes required to transmit a frame over the wireless channel.

The implemented MAC is available as open-source code for the research community at the following site.