An example of the usage of wireless LAN technologyfor multimedia: see Berlin live!

TKN - Telecommunication Networks Group TU-Berlin
Head of Group: Prof. Adam Wolisz Faculty of Electrical Engineering and Computer Science


Main Research Areas within the AMICA Architecture

We follow the vision of end-systems being connected mainly via wireless technologies to an optical communication infrastructure. We believe that the communication will be based on multimedia- support enhanced Internet, with quality of service support being achieved by segregation rather than reservation.
In addition to wireless access also nomadicity and mobility will have to be increasingly supported. In our approach, called AMICA: Adaptable, Mobile; Internet-based Communication Architecture we investigate several crucial aspects of such future systems, following original research hypothesis.

  • Transport Protocols services for networks with wireless last hop(s).

    Efficiency problems of TCP over wireless are well known.
    We believe that the data flow in the internet
    "backbone - underlying the global congestion control rules"
    should be decoupled from the data pacing on wireless
    hop, which should be locally optimized.
    We follow the idea of supporting TCP/UDP
    sockets (with original semantics) in a proxy-based
    architecture called ReSoA (Remote Socket Architecture).
    This architecture consists of two Layers:

    • The socket export Layer, which supports the remote access

      to the transport Layer services, and is dependent only on
      their semantics. Thus this layer should be changed only if,
      for example, new socket interface would be suggested.

    • The link layer protocol, called Last Hop Protocol (LHP)
      offering different Quality of Service to different
      flows (up to -possibly - knowledge of the semantic
      of the coding schema for voice, video,..etc)
      The following of the ReSoA approach, makes it much easier
      to extract the information about the individual flows
      (e.g. using the port numbering convention) without the
      need of decoding IP/UDP packet headers or extending
      IP protocol fields.

    Following the ReSoA approach- with congestion control for the
    backbone located in ReSoA server not in the end-system will
    also make it much easier to enforce globally unifies congestion
    control - also for UDP traffic.

    We develop a formal specification, simulation models
    and prototype implementation for this architecture.
    For comparison of its performance with other approaches
    (Snoop, Eiffel,,etc) we develop a unified simulation and
    experimental environment

  • Link Layer Protocols for transmission optimization over wireless links
    Wireless links are known for varying quality.
    We believe that efficient support of diffferent applications
    might be best achieved by link layer protocols which use both:

    • Statistical knowledge about the channel, supported by
      actual dynamic knowledge of the temporary state
      of the channel

    • Use of information about the semantic of individual
      data flow (up to -possibly - knowledge of the semantic
      of the coding schema for voice, video,..etc)

    We work on approaches using the fluctuation of the channel
    quality, for non-uniform, or even non-continuous transmission
    rather than trying to achieve constant link quality. We develop
    a family of Link-Layer protocols for different assumptions on
    Physical Layer: IEEE802.11 type, CDMA Type  etc. as well as
    for different flows: WWW_traffic, MPEG Video, Voice over IP...

    One good example of our activities is the usage of multiple CDMA
    codes in parallel in good channel states to remove the backlog
    of data packets created during the fading phase in CDMA based
    systems (like UMTS).

  • Usage of Multiple hops in the wireless access

    Due to fundamental laws of Physics the energy needed for
    successful transmission is proportional to the distance
    in the power of 3 or 4. Thus - theoretically- usage of
    multiple hops should decrease the needed energy, as well as
    the interference generated by the senders involved.
    On the other hand processing in intermediate hops causes
    energy usage and delays, hidden terminal effects appear..
    We investigate the potential and  possible solutions for
    increasing system capacity and end-system energy usage by
    introducing multi-hop, ad-hoc organization of traffic among
    end-systems and base stations in cellular systems.

  • MAC Protocols for wireless LANs

    We believe that an important role within the plethora of
    different wireless technologies will be played by Wireless
    LANs. Therefore we investigate the performance, as well
    as possible modifications of MAC protocols, as well as the
    schedulers implemented of top of them. We consider different
    technologies: IEEE 802.11 being the most intensively studied,
    but also HiperLAN and Bluetooth.
    We are interested in development of scheduling rules and
    protocols supporting different types of traffic flows.
    For example one specific class of traffic, which we would
    like to support with proper MAC on top of IEEE802.11 PHY
    is industrial process control traffic, we aim at offering
    in wireless a service semantics similar to that of PROFIBUS.
    We are also interested in exploiting in MAC the features of
    new Physical Layer technologies, like OFDM and smart antennas.

  • Handover in Heterogeneous Networks

    There seems to be more and more interest in a vision of
    hierarchical radio coverage using different radio technologies
    from different radio providers rather than a unique,
    homogeneous infrastructure, like in today cellular networks.
    Handover in the heterogeneous environment means change of the
    Physical layer technology to one with different Quality of Service,
    means possibly a dramatic change of the information routing (e.g.
    change form the WLAN extension of the campus network to an
    overlay from an INT-2000/UMTS operator) as well as authorization
    /charging/billing schemata.
    We believe that the basic Mobile IP approach, designed for
    nomadicity support is not enough for the future mobile Internet.
    We believe that there will be a plethora of approaches, assuring
    mobility in different scenarios and at different levels.
    IN our research we follow selected ideas. We investigate the
    possibility of using multicast as a basic mechanism for handover
    support, and consider which additional features should a future
    multicast mechanism have in order to offer optimal handover
    support (MOMBASA). We believe that the development of handover functionalities
    so far mostly ignored the existence of Performance enhancing
    proxies and investigate approaches to unify the handover enhancing
    proxies with PEPs. We focus on perfomance and security aspects
    of different handover mechanisms and policies.

  • Protocol Reloading/Modification in Terminals and Access Points

    Switching among different technologies needs not only change of the
    Physical layer, which will be in the future assured by soft-radio
    and reconfigurable radio technologies, but also change of the
    MAC/Link Layer protocols and possibly deployment of proxy clients
    in the end-systems.
    We believe that this should be done in a dynamic way, by downloading
    proper modules to the end-systems rather than pre-storing them.
    We are investigating the execution environments for end-systems as
    well as secure and efficient download mechanisms.
    On the other hand, the proxy servers - both for mobility and
    performance enhancing proxies-  have to be deployed at access points
    and other infrastructure elements. We are investigating the potential
    of current active network concepts to support such deployment, we
    are interested in defining the minimum functionality of active nodes
    needed for such deployment

  • Power Saving in Mobile Networks

    Power economy is definitely one of the important design criteria
    for wireless communication support of numerous classes of end-systems:
    lap-tops, palm-tops but also sensors.
    We believe that an impressive energy economy effect might be achieved
    by proper tuning of the protocol stack. Typical actions which we have
    investigated for W-LANs, are optimization of active/sleep phases of
    mobile hosts, but also joint optimization of packet length and
    transmission power, possibly in pace with the dynamics of the channel
    quality. We are interested in harmonization and joint optimization of
    operation of the whole protocol stack from the power economy point of view.

  • Admission Control and Route Selection in Access Networks

    Despite the increasing capacities of the fixed network, the
    proper Quality of Service in fixed network can not always be
    expected, especially for more demanding real-time traffic.
    We believe that traffic separation in backbone is the proper way,
    route separation being one of promising approaches. We investigate
    measurement based decision policies as for route selection and
    admission control.

  • IP over WDM

    We believe in an "optical transmission meets wireless
    transmission" future of the Internet. Therefore in
    addition to supporting the vision of Wireless/mobile
    internet access, we investigate the options for
    "IP over WDM". We follow here especially the idea to
    use AWGs in order to create both single-hop and
    multi-hop architectures supporting IP over WDM.
    We are also interested in network structures for
    efficient use of the "radio on the fiber concept"

Questions? Contact webmaster.

Contents subject to change. All rights reserved.
Mit dem Urteil vom 12. Mai 1998- 312 O 85/98- "Haftung für Links" hat das Landgericht Hamburg entschieden, daß man durch die Anbringung eines Links, die Inhalte der gelinkten Seite ggf. mit zu verantworten hat. Dies kann nur dadurch verhindert werden, daß man sich ausdrücklich von diesen Inhalten distanziert.
"Hiermit distanzieren wir uns ausdrücklich von allen Inhalten aller extern gelinkten Seiten auf unserem Server und machen uns diese Inhalte nicht zu eigen. Diese Erklärung gilt für alle auf unserem Server angebrachten externen Links."