Back-Pressure-Based Packet-by-Packet Adaptive          Routing in Communication NetworksAbstract—      Back-pressure-based...
pressure routing algorithm in this paper and design a new algorithm that has muchsuperior performance and low implementati...
new idea here is to perform routing via probabilistic splitting, which allows thedramatic reduction in the number of real ...
back-pressure algorithm, is used to activate transmissions between nodes.       However, first, actual transmissions send ...
Upcoming SlideShare
Loading in …5
×

Back pressure based packet by packet adaptive

1,322 views

Published on

For further details contact:

N.RAJASEKARAN B.E M.S 9841091117,9840103301.


IMPULSE TECHNOLOGIES,
Old No 251, New No 304,
2nd Floor,
Arcot road ,
Vadapalani ,
Chennai-26.

0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
1,322
On SlideShare
0
From Embeds
0
Number of Embeds
1
Actions
Shares
0
Downloads
59
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Back pressure based packet by packet adaptive

  1. 1. Back-Pressure-Based Packet-by-Packet Adaptive Routing in Communication NetworksAbstract— Back-pressure-based adaptive routing algorithms where each packet isrouted along a possibly different path have been extensively studied in theliterature. However, such algorithms typically result in poor delay performance andinvolve high implementation complexity. In this paper, we develop a new adaptiverouting algorithm built upon the widely studied back-pressure algorithm. Wedecouple the routing and scheduling components of the algorithm by designing aprobabilistic routing table that is used to route packets to per-destination queues.The scheduling decisions in the case of wireless networks are made using counterscalled shadow queues. The results are also extended to the case of networks thatemploy simple forms of network coding. In that case, our algorithm provides alow-complexity solution to optimally exploit the routing–coding tradeoff.Reasons for the Proposal: THE BACK-PRESSURE algorithm introduced in has been widely studiedin the literature. While the ideas behind scheduling using the weights suggested inthat paper have been successful in practice in base stations and routers, theadaptive routing algorithm is rarely used. The main reason for this is that therouting algorithm can lead to poor delay performance due to routing loops.Additionally, the implementation of the Back-pressure algorithm requireseachnode to maintain per-destination queues that can be burdensome for a wirelineor wireless router. Motivated by these considerations, we reexamine the back
  2. 2. pressure routing algorithm in this paper and design a new algorithm that has muchsuperior performance and low implementation complexity.Existing System: Prior work in this area has recognized the importance of doing shortest pathrouting to improve delay performance and modified the back-pressure algorithm tobias it toward taking shortest-hop routes. A part of our algorithm has similarmotivating ideas. In addition to provably throughput-optimal routing thatminimizes the number of hops taken by packets in the network, we decouple (to acertain degree) routing and scheduling in the network through the use ofprobabilistic routing tables and the so-called shadow queues. The min-hop routingidea was studied first in a conference paper, and shadow queues were introduced inhand, but the key step of partially decoupling the routing and scheduling whichleads to both significant delay reduction and the use of per-next hop queuing isoriginal here. In the authors introduced the shadow queue to solve a fixed routingproblem. The min-hop routing idea is also studied in, but their solution requireseven more queues than the original back-pressure algorithm. Compared to, themain purpose of this paper is to study if the shadow queue approach extends to thecase of scheduling and routing. The first contribution is to come up with aformulation where the number of hops is minimized. It is interesting to contrastthis contribution with. The formulation in has the same objective as ours, but theirsolution involves per-hop queues, which dramatically increases the number ofqueues, even compared to the back-pressure algorithm. Our solution issignificantly different: We use the same number of shadow queues as the back-pressure algorithm, but the number of real queues is very small (per neighbor). The
  3. 3. new idea here is to perform routing via probabilistic splitting, which allows thedramatic reduction in the number of real queues. Finally, an important observationin this paper, not found in, is that the partial ”decoupling” of shadow back-pressureand real packet transmission allows us to activate more links than a regular back-pressure algorithm would. This idea appears to be essential to reduce delays in therouting case, as shown in the simulations.Proposed System: We also consider networks where simple forms of network coding areallowed [7]. In such networks, a relay between two other nodes XORs packets andbroadcasts them to decrease the number of transmissions. There is a tradeoffbetween choosing long routes to possibly increase network coding opportunities(see the notion of reverse carpooling in [8]) and choosing short routes to reduceresource usage. Our adaptive routing algorithm can be modified to automaticallyrealize this tradeoff with good delay performance. In addition, network codingrequires each node to maintain more queues [9], and our routing solution at leastreduces the number of queues to be maintained for routing purposes, thus partiallymitigating the problem. An offline algorithm for optimally computing the routing–coding tradeoff was proposed in [10]. Our optimization formulation bearssimilarities to this work, but our main focus is on designing low-delay onlinealgorithms. Back-pressure solutions to network coding problems have also beenstudied in [11]–[13], but the adaptive routing–coding tradeoff solution that wepropose here has not been studied previously. We summarize our main results asfollows. • Using the concept of shadow queues, we partially decouple routing and scheduling. A shadow network is used to update a probabilistic routing table that packets use upon arrival at a node. The same shadow network, with
  4. 4. back-pressure algorithm, is used to activate transmissions between nodes. However, first, actual transmissions send packets from first-in–first-out (FIFO) per-link queues, and second, potentially more links are activated, in addition to those activated by the shadow algorithm. • The routing algorithm is designed to minimize the average number of hops used by packets in the network. This idea, along with the scheduling/routing decoupling, leads to delay reduction compared with the traditional back pressure algorithm. • Each node has to maintain counters, called shadow queues, per destination. This is very similar to the idea of maintaining a routing table per destination. However, the real queues at each node are per-next-hop queues in the case of networks that do not employ network coding. When network coding is employed, per-previous-hop queues may also be necessary, but this is a requirement imposed by network coding, not by our algorithm. • The algorithm can be applied to wireline and wireless networks. Extensive simulations show dramatic improvement in delay performance compared to the back-pressure algorithm. The rest of this paper is organized as follows. We present the network model in Section II. In Sections III and IV, the traditional back-pressure algorithm and its modified version are introduced.We develop our adaptive routing and scheduling algorithm for wireline andwireless networks with and without network coding in Sections V–VII. In SectionVIII, the simulation results are presented

×