Vous êtes sur la page 1sur 5

International Journal of Computer Trends and Technology- volume3Issue1- 2012

Capacity (Cut-Set) Bounds for Multi-Hop Relay Networks in Erasure Channels


B. Sree deepthi#1, K. Ujwal Deep*2 and G .Mary silvya*3
M.tech Student Department of ECE,K L University Andra Pradesh, India. *2,3 Asst.Professor Department of ECE,K L University Andra Pradesh, India.
#1

Abstract This paper deals with the finding of capacity bounds for multi-hop relay networks in erasure channels. As importance of complex networks of nodes and relays and significance of erasure channels are well identified, this paper gives the way of finding upper and lower capacity bounds by using mutual information concept with respect to information theory aspects. Keywords Erasure channel, multi-hop, cut-set bound, mutual information, information theory.

I. INTRODUCTION A multi-relay network is considered in which communication from source to relays takes place over a channel, the channel connecting these nodes is supposed to be an erasure channel where symbols are received correctly without any error, or lost due to buffer overflows or excessive delays. The relay forward some part of the information and we assume that relay and sender transmission do not interfere with each other, however the receiver is able to receive in parallel information that are sent from sender and relay on different channels. The interference between sender and relay transmission might be suppressed by using different sender to receiver and relay to receiver physical channels. Erasure channels are chosen as the work can be continued easily to another channels basing on this.

ri(2), ri (h) = di. In time slot t+1 we have nodes r1(t), r2(t).... rk(t) transmitting simultaneously to nodes r1(t+1), r1(t+2),,rk(t+1) respectively. Nodes r1(t+1), r1(t+2),,rk(t+1) decode their respective signals x1; x2;.; xk and transmit them to the next set of relay nodes in the (t + 2)th time slot, and so on. A natural condition to impose is that the relay nodes that are receiving (or transmitting) messages in any time slot be distinct so that the messages do not collide. In addition, impose the constraint that relay nodes cannot receive and transmit at the same time. In the rest of the chapter, refer to these conditions together as the no collisions property. In general, ri(t) = ri(t+1) for any value of i. This means that a relay can effectively hold on to a message in a time slot; hence h effectively represents the maximum number of hops needed for all the source-destination pairs.

1-pe 1
pe

1 ? 0 Y
Figure 2. Multi-Hop Relay Network in Erasure Channels. II. CAPACITY BOUNDS This section will present theoretical bounds for the capacity of the relay channel case. We first begin with the simple case of single relay and we extend the analysis to multiple relay situations.

X 0

pe

1-pe
Figure 1. Binary Erasure Channel with Erasure Probability pe .

In general, suppose that the source-destination pair (si; di) communicates using a sequence of relay nodes ri. (h = 1; 2; represents the number of hops.) Denote ri(0) = A. Single Relay Case si , ri(h) = di. The path from si to di is then ri(0) = si , ri(1),

ISSN: 2231-2803

http://www.internationaljournalssrg.org

Page 29

International Journal of Computer Trends and Technology- volume3Issue1- 2012


In this section we consider the simplest case of an erasure relay channel which is a network composed of one sender, one receiver and one intermediate or relay node (Fig. 3). In this case, the relay channel can be described with five random variables X0, X1,Y0,Y1 and Y2 and a conditional probability density function p ( y 0 , y1 , y 2|x, x1 ) . This last function gives the probability that when X0 is sent by sender and X1 by the relay, Y1and Y2 are received at the receiver and the relay receives y1. We further define Y = (Y1, Y2 ) as the received variable from sender and relay at the receiver. Lets consider a special case of relay network where the relay channel consists of two separate and independent (non interfering) erasure broadcast channels: from sender to all relays and the receiver ((X 0 ; Y 0, Y 1 )) and from relay to all other relays and the receiver (X 1 ; Y 2 ,) (in the single relay case this broadcast channel is only point to point). We further assume that the erasure probabilities are p, p1 and p2, where p, p1 and p2 are the packet loss rate of sender-receiver, sender-relay and relay-receiver links (see Figure3).we a ss u me t h at no collision occurs between the relay to receiver and the sender to receiver transmissions. This separation can be achieved through using different physical channels, for example by making the sender operating in frequency f1 and the relay sending over frequency f 2 . Moreover if the sender to relay channel is more capable than the sender to receiver channel, i.e. I(X0; Y0) <I(X0; Y1), the bound is simplified as : R = sup min{ I(X0; Y1) I(X1; Y1), I(X0; Y0) + I(X1; Y2|X0)} p(x1,x2) Theorem 2: The capacity region over an erasure relay channel is bounded as below[5]: R max min{(1-p.p1),(1-p)+(1-p2)} Under the situation that the sender to relay channel is more capable than the sender to receiver channel the bound can be tightened and become: R max min{(1-p1),(1-p)+(1-p2)} where p1 is the loss probability between sender and relay, p2 is the loss probability between relay and receiver and p is the loss probability between sender and receiver. is a coupling parameter 0 1. B. Multi-Relay Case The single relay case presented in previous section can be extended in a straightforward way to the general case of more than one intermediate node. Consider a set of N node N = {1, . . . ,N} communicating over a general erasure channel characterized by a loss matrix P = (pi,j)NxN , where pi,j is the loss probability of a transmission between node i and j. Now lets suppose that there is a single communication taking place between a single sender-destination pair (s, d). Lets suppose that nodes in N are partitioned into M + 1 disjoint level sets (L0,L1, ...,LM,LM+1) where L0 = {s} is the sender node and LM+1 = {d} is the destination node. We assume that the nodes in Lm receive information only from the nodes in previous levels Lm-k for k [1, m], while they send information only to nodes in the next levels Lm+k for k [1,M+1-m] . Lets nm be the number of nodes in level Lm. By definition n0 = nM+1 = 1. Lets Xkm,j denote the message sent by node Lm,j (node j in layer m) at time k and Xkm = (Xkm,1,Xkm,2, . . . ,Xkm,nm) be the vector of message sent by layer m at time k. Now let Y k m,j be the message reveive by node j in layer m at time k. We assume here that messages Xk m will only be received at time k by receivers in layer l > m(see fig.4). This bound should be compared with the maximum rate achievable by classical routing based relaying system, where a packet is sent to the relay and directly forwarded to the receiver. In this case the achievable rate will be bounded by R min {(1 p1), (1 p2)}. The obtained bounds in theorem 3 are clearly larger than the routed based bound for the same scenario, but as we will see reaching these bounds come at the cost of a network coding (a processing cost at

Figure 3. Relay Channel The cut-set bound is the most general bound that can be defined in the context of relay channels. This bound can be expressed for the relay channel described in figure 3 as: R sup min{I (X 0 ; Y 0 , Y 1|X 1 ), I (X 0 , X 1 ; Y 0 , Y 2 )} p(x1 ,x2 ) The capacity region bound of erasure relay channel with an intermediate node is determined by the following theorem. Theorem 1 (Capacity region bound): The capacity region of the relay channel in Fig.1 is bounded by R = sup min{ I(X 0; Y0, Y1) I(X1; Y0, Y1), p(x1,x2) I(X0; Y0) + I(X1; Y2|X0)}

ISSN: 2231-2803

http://www.internationaljournalssrg.org

Page 30

International Journal of Computer Trends and Technology- volume3Issue1- 2012


each router) for retrieving information in X0 and enabling the collaboration through sending an independent but relevant information X1 where the maximization occurs over all possible relay load for each relay in the network. Moreover Rk, k = 1, . . . ,M is defined as k-1 k k = min {(1 p ,k ) + p )} R 01 i hj (1- hj,ki i=1,..,nk n-1 n-1 where nk is the number of nodes in the level Lk , hj is the relay load carried by the relay node h in level Lj and phj,ki is the loss probability from node j in layer h to node i in layer k. And the above expression gives the value for upper cut-set bound for multi-hop relay networks in erasure channels. C. Lower (Cut-Set) Capacity Bound: The value of lower (cut-set) capacity bound for multihop relay networks in erasure channel is given by R max min { I(X0; Y 0, Y 2|X2)} and in terms of erasure probabilities, the above expression becomes

Figure 4. Multi relay channel, level Lm has nm nodes for m [1,M].L0 = {s} and LM+1 = {R}.

When p1 < p2 the capacity bound is governed by p2 and (1 p2) is an upper bound of the capacity. In this case it worth to send all the information through the relay channel, i.e. I(X0; Y 0, Y 1|X1) = (1 p.p1)H(X0) the relay will have to transmit all the received information. when p2 > p it this case the capacity region is bounded by 1 R max min (1-p.p2) p and the relay is useless as the information dispersion resulting from the relaying schemes will harm more than it where is the parameter, which tells about the amount of benefits the overall rate. information carried by the intermediate relays. And its value is Using the same approach that the single erasure relay always less than or equal to the value of . case and using the Shearer theorem, under the degraded hypothesis, this bound can be simplified for the erasure channel: Theorem 3(Capacity bounds for erasure multi-relay case)[7]: The capacity region of a degraded erasure multi relay channel is bounded by R < max [min{R1,R2, ...,RM}] RM < I (XM ; YM Where R1= min I (Xm; Ym+1,j |Xm+1, . . . Xm ),
j [1,n m + 1 ] +1

III. SIMULATED RESULTS Obtained capacity bounds values are simulated using MATLAB software taking erasure probabilities on y label and capacity (cut-set) bound values on x label and graph is plotted by considering different values in different scenarios.

Rm < min{R1 , Rm+1 + R2 }

R2 = min I (Xm; Ym+k,j |Xm+1, . . . , XM }


j [1,n m + 1 ],k[2,M +1m]

for m = M 1, M 2, . . . , 0.

ISSN: 2231-2803

http://www.internationaljournalssrg.org

Page 31

International Journal of Computer Trends and Technology- volume3Issue1- 2012


Figure 5. In this case p1=0.7; p2=0.9; alpha=0.1; p varies from 0 to 1

Figure 9. In this case p1, p varies from 0 to 1; alpha=0.3; p2= 0.1

Figure 6. In this case p1, p2= varies from 0 to 1; alpha=0.1; p= 0.2

Figure 8. In this case p1= 0.3, p2=0.7; alpha=0.9; p= varies from 0 to 1

Figure 7. In this case p1, p2= varies from 0 to 1; alpha=0.5; p= 0.9 In future, this work can be extended for deriving the capacity bounds for different channels, and for multi-access scenarios using single cut-set bound or multiple cut-set bounds in entire the network.

IV. CONCLUSION To transmit information nearly without error at any rate below a limiting rate, in communication theory, Shannons capacity theorem yields good results. But it was restricted to point-to-point communications. And when we are dealing with network of nodes it was a problematic issue to calculate the capacity bounds so mutual information concept with respect to information theory aspects is used to derive the capacity bounds of multi-hop relay networks in erasure channels.

ACKNOWLEDGMENT The authors express their thanks to the management of KL University and the Department of Electronics and Communication Engineering for their support during this work.

ISSN: 2231-2803

http://www.internationaljournalssrg.org

Page 32

International Journal of Computer Trends and Technology- volume3Issue1- 2012


REFERENCES [1] T. M. Cover and J. A. Thomas. Elements of Information Theory. John Wiley Sons, New York, 1991. [2] Piyush Gupta and P. R. Kumar.The capacity of wireless network. IEEE. Trans. Inform. Theory, vol. IT-46(no. 2):pp. 388404, March 2000. [3] L.-L. Xie and P. R. Kumar, A network information theory for wireless communication: Scaling laws and optimal operation, IEEE Trans. Inform. Theory, submitted April 2002. [4] P. Gupta and P. R. Kumar. towards an information theory of large networks: An achievable rate region. IEEE. Trans. Inform. Theory, vol. IT-49:pp. 18771894, August 2003. [5] R Khalili.On the achievability of cut-set bound for a class of erasure relay channels -The non degraded case.ISITA conference, 2004. [6] R. Khalili and K. Salamatian. a new relaying scheme for cheap wireless relay nodes. wiopt, 2005. [7] G. Kramer, M. Gastpar, and P. Gupta, Cooperative strategies and capacity theorems for relay networks, IEEE Trans. Inf. Theory, vol. 51, no. 9, pp. 30373063, Sep. 2005. [8] K. Salamatian and R. Khalili, An information theory for the erasure channel, in Proc. of the 43rd Annual Allerton Conference on Communication, Control, and Computing, Allerton, IL, Sep. 2005. [9] R. Khalili and K. Salamatian, A tighter cut-set bound for the multi-terminal erasure channel without side information, in Proc. IEEE Int. Symposium on Inform. Theory, Seattle, WA, pp. 18761880, Jul. 2006. [10] L.-L. Xie and P. R. Kumar, Multisource, multidestination, multirelay wireless networks, IEEE Trans. Inf. Theory, vol. 53, no. 10, pp. 35863595, Oct. 2007. [11] M. A. Khojastepour, A. Sabharwal, and B. Aazhang. bounds on achievable rates for general multi-terminal networks with practical constraints . LNCS, April 2003. [12] L. Sankaranarayanan, G. Kramer, and N. B. Mandayam, Capacity theorems for the multiple-access relay channel, in Proc. of the 42nd Annual Allerton Conference on Communication, Control, and Computing, Allerton, IL, pp. 17821791, Sep. 2004. [13] A. Sanderovich, S. Shamai (Shitz), Y. Steinberg and G. Kramer, Communication via decentralized processing, IEEE Trans. Inform. Theory, vol. 54, no. 7, pp. 30083023, Jul. 2008. [14] Kurniawan, E.; Sun, S.; Yen, K.; Transmission strategy of fountain code in cooperative networks with multiple relay nodes,in proc. IEEE,2009. [15] Q Yan, DW Yue Cooperative Diversity of Wireless Networks with Multiple Amplify-andForward Relays and Hard-Decision Detections,in proc. of mobile computing,CMC-IEEE,2010.

ISSN: 2231-2803

http://www.internationaljournalssrg.org

Page 33

Vous aimerez peut-être aussi