Wired vs Wireless Discussion

1
Wired vs WirelessDiscussion
2
Wired vs Wireless

• The distinction between these networks is
  definitely becoming less and less marked, and to
  an extent, network protocols and architectures
  may also have to move in that direction.
• Anything that will call itself "general
  networking" must necessarily include wireless
  networking
• There is still a big difference between wired and
  wireless networks in network/transport
  performance beyond raw link data rates

3
Wired vs Wireless

• Wireless offers a different value proposition
  from wired networks more about mobility and
  freedom and less about performance. Wireless
  enables new unique applications, so that users
  are willing to pay more per bps.
• The forthcoming explosion of wireless/mobile
  devices should drive the design of the global
  Internet in terms of core features such as
  naming, addressing, routing, content/location
  awareness, and security.
• There is an opportunity to merge, but this can
  only happen if there is cross-fertilization
  between communities.

4
Wired vs Wireless

• Wireless is a very good access technology and
  last-resort long-haul technology, while wireline
  is a great long-haul technology and a good access
  technology and there's no reason to believe the
  broad respective merits of the two classes of
  technology will change in the near future.
• Wireless networks must be considered to be part
  of the network because that is the user,
  management, security, enterprise, economic, and
  perhaps regulatory model that exists, unless one
  believes in completely unwired global networks.
  
• Most new requirements of future networking will
  come out of the mobile and wireless area and goes
  beyond what is need only for wired networks

5
Areas of Dichotomy

• Errors, Power and Topology
• We don't think power is as much a dichotomy as
  has been suggested.
• The main differences are in the technologies and
  protocols for aggregation in the core vs.
  efficient delivery in the access network,
  but other requirements are more alike than
  commonly believed.  

6
Areas of Dichotomy

• Both technologies use some part of the
  spectrum,in some media, to transmit and receive
  data. But the nature of the respective spectrums
  is wildly different.
• Wired networks are more established, and thus the
  parts within the cloud are harder to modify
  incrementally.

7
Areas of Dichotomy

• The research challenges lie in the fact that
  wireless networks are very niche and to an extent
  may prevent the adoption of more generalized
  concepts.
• Two examples are traffic loss, and management of
  some kinds of resources very different, e.g.
  power.
• A qualitatively different dichotomy is reflected
  in the the underlying broadcast nature of
  wireless that could be used to provide very
  different communications paradigms much more
  naturally than wired

8
Areas of Dichotomy

• Security, Naming and Mobility, routing and
  quality assurance, Management
• Administration-free, "self-managing" paradigm for
  ad hoc naming and management of personal devices
  regardless of whether they are attached via
  wireless or wired networking technologies.
• This is largely the wrong list for wireless --
  most of the wireless research I'm at least
  peripherally involved with fits in none of these
  bins comfortably

9
Areas of Dichotomy

• Security In wired networks, fingerprinting can
  be foiled by topology/firewalls, but these aren't
  available to first/last hop wireless
• Each new protocol opens avenues for attack that
  can magnify attacker effort
• Session continuity
• Media delivery and adaptation 

10
Dichotomies into the future

• Routing IPv4 and IPv6
• While layering has permitted great strides in the
  past, it has also caused networking research to
  be performed in an application agnostic manner.
  Perhaps this is hampering the next big revolution
  in networking
• Integrating wireless networks and mobile users
  efficiently will require greater awareness of
  network conditions and disconnected/opportunistic
  modes that currently don't appear in the CDN model

11
Dichotomies into the future

• Handling errors in wireless remains an open
  issue, especially when performance-enhancing
  proxies aren't an option
• Different layers of the network are used to using
  different approaches, and this can cause
  contention.
• Since the user has a single, perhaps
  long-distance experience built on a composition
  of underlying resources, both wired and wireless
  infrastructures, to truly effectively manage
  this, information and/or analysis will need to
  flow across that wired/wireless boundary

12
Dichotomies into the future

• A consolidated QoS/Security/Mobility
  framework/solution (all through the stack,
  possibly except for the application level)
• Dynamic and adjustable content delivery