Energy accumulation in destination nodes of wireless relay networks

Abstract

A method for transmitting packets from a source node to a destination node via relay nodes of a wireless network. Packets are transmitted from the source node, along a route of relays nodes, to the destination node in a wireless network. Energy of the packets is accumulated only in the destination node by storing multiple versions of the packet. The packets are decoded in the destination node using the accumulated energy.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a block diagram of a wireless relay network according to an embodiment of the invention;

FIGS. 2A-2D are block diagrams of a network with additional relay nodes;

FIG. 3 is a block diagram of a data packet and a request to cooperate packet (RTC) according to an embodiment of the invention;

FIG. 4 is a block diagram of descriptions of fields in the RTC packet of FIG. 3 according to an embodiment of the invention;

FIG. 5 is a block diagram of pseudo-code executed by a relay node of the network of FIG. 1 according to an embodiment of the invention; and

FIG. 6 is block diagram of pseudo-code executed by other nodes receiving a packet according to an embodiment of the invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT

Wireless Relay Network

FIG. 1 shows a wireless relay network 100 according to an embodiment of our invention. The network includes a source node s 111, a destination node t 131, and one or more intermediate decode-and-forward relay nodes r 121-124. All nodes use unicast transmission via single omni-directional antennas for transmission and reception, and operate in half-duplex mode, i.e., the nodes can either transmit or receive, but not do both simultaneously. The network 100 is quasi-static, in which occasional link updates reflect possible changes of channel state information of channels of the network. The source can transmit directly to the destination, or indirectly via one or more relay nodes. The relay nodes can forward packets to the destination serially or in parallel.

Destination Energy Accumulation

The embodiments of the invention use destination energy accumulation (DEA). DEA fills the gap between the two known extremes, namely (i) a conventional network, which requires simple decode-and-forward relays that do not benefit from energy accumulation, and (ii) a complete energy-accumulation network, which requires highly complex decode-and-forward relays that accumulate energy to the greatest possible extent.

In our embodiments, only the destination node uses multiple stored versions of the packet to decode the packet, while an intermediate relays does not store multiple versions of a packet. That is, the relay nodes discard the packet after the packet has been forwarded. In one of the embodiment, versions of the packet are copies of the packet.

A cyclic redundancy check can be inserted in the packet to determine whether the packet is decoded correctly. Energy accumulation only at the destination is justifiable for the following reasons. By energy accumulation, we specifically mean storing multiple versions of the same packet only at the destination node. In many sensor networks, the destination node, which typically gathers sensor data from all sensor nodes, usually has greater computational, memory and power resources. In addition, the effort of accumulation occurs at the node that benefits from the accumulation. The number of packets that need to be accumulated and stored is limited. Furthermore, energy accumulation only at the destination reduces energy consumption throughout the network. As another advantage, energy accumulation only at the destination significantly simplifies route discovery, and makes a practical implementation feasible.

Progressive Accumulative Routing

We also use a progressive accumulative routing (PAR) process, which determines an energy efficient DEA route, and sets the node transmit powers in a distributed and progressive manner. As a distributed process, PAR establishes energy efficient accumulative routes based only on local channel state information available at each node. The progressive nature of the process enables incremental addition of new nodes to an established DEA route, and realizes additional energy reductions.

Due to changes in the propagation environment or due to the mobility of the nodes, the channels between the various nodes changes with time. PAR can be used to update an already established route.

The PAR process significantly improves the total energy efficiency compared to conventional non-accumulative networks. That is, the amount of energy that is consumed while transmitting packets along the route is decreased. With a high probability, the PAR process performs as well as optimal complete energy accumulation at all nodes.

Network Model

Let V be the set of nodes in the network 100. For nodes U, v ε V, let h_uv be the absolute value of the channel gain between node u and node v. A node can only determine its channel gain with respect to neighboring nodes. The node need not determine the phase of any channel gain, nor can the node determine any other gain of links between other nodes.

A node can forward a packet only after having reliably decoded that packet. According to an embodiment of the invention, only the destination node accumulates energy by storing multiple versions of the packet, while the relay nodes do not. The destination node can receive and store multiple “soft” versions of the same packet from multiple nodes.

The packet can be successfully decoded by the destination node after the total energy accumulated from the multiple received versions of the packet exceeds a predetermined threshold, which depends on a modulation and a coding used for transmission, see Maric et al., and. Agarwal et al., above, incorporated herein by reference. A cyclic redundancy check (CRC) may be included in the packet to enable the receiver to determine if it has correctly decoded the packet or not.

If the destination receives one version of the packet from each of nodes u₁, u₂, . . . , u_n, then the destination can decode the packet successfully when the total accumulated power

$\sum _{k=1}^np_kh_{u_kt}$

is equal or greater than the threshold γ, where p_k is the transmit power of node u_k. A relay node v can successfully decode the packet transmitted by node u with power p if and only if ph_uv≧ γ, otherwise, the relay discards the undecodable packet. Without loss of generality, a duration of a packet is normalized to unity. Therefore, we interchangeably use the terms energy and power.

Progressive Accumulative Routing

We consider a single source, s, and a single destination, t. First, we derive the general conditions for power reduction when (i) a single relay is added between the nodes s and t, and (ii) when a second relay is introduced in an energy accumulative route that already includes one relay. As described below, very limited information is often needed to determine the optimal relay. Then, we extend the result to a general energy accumulative route that includes an arbitrary number of relays. We also describe how additional energy reduction can be achieved using the local channel state information at the relays and limited additional information.

Adding a First Relay Between the Source and the Destination

Lemma 1

An accumulative route from the source node s to the destination node t through relay node r can reduce a total power consumption if and only if there exists a node r, such that

h_st<min{h_sr, h_rt}  (1)

The maximum total power reduction, P_s^red(r), by having node r act as a relay is given by

P _s ^red(r)=(1−h_st/h_sr)(1−h_st/h_rt)( γ/h_st),  (2)

and is achieved when nodes s and r set their transmission powers P_s and P_r, respectively, at

P _s=(1/h_sr) γ, and P_r=(1/h_rt)(1−h_st/h_sr) γ.  (3)

Proof

First, we assume that none of the nodes satisfy equation (1). This implies that h_st≧h_sr, and/or h_st≧h_rt, for all relays r ε V−{s,t}. For any node, r, if h_st≧h_sr, then less power is required transmit a packet successfully to the destination than to the relay. If h_st≧h_rt, given the same transmission power, the destination receives a higher signal power if a packet is transmitted by the source and not the relay. Hence, the use of a relay cannot reduce the total power consumption.

Let there exist at least one node, r, such that h_st<min{h_sr, h_rt}. In DEA, if r is a relay, then the source first transmits a packet with power P_s so that node r can decode the packet successfully. Then, node r transmits the packet to the destination node t with power P_r. The destination decodes the packet using the energy accumulated from the transmissions of both nodes s and r. Hence, the optimal power allocation problem is the following:

$\begin{array}{cc}\underset{P_s,P_r}{\min }P_s+P_r\mathrm{subject}\mathrm{to}\left[\begin{array}{cc}{h}_{\mathrm{sr}}& 0\\ h_{\mathrm{st}}& h_{\mathrm{rt}}\end{array}\right]\left[\begin{array}{c}{P}_s\\ P_r\end{array}\right]\ge \left[\begin{array}{c}\stackrel{\_}{\gamma }\\ \stackrel{\_}{\gamma }\end{array}\right].& \left(4\right)\end{array}$

The first inequality in the constraint in equation (4) ensures that node r decodes the packet transmitted by node s. After node r decodes the packet, it is more energy efficient to let node r deliver the remaining energy for node t to decode the packet, because h_rt>h_st. This leads to the power allocation in equation (3), which satisfies the constraint in equation (4) with equality. The total power reduction with the power setting in equation (3), compared to the minimum power, γ/h_st, required for a direct transmission from node s to node t, is then given by equation (2). This power reduction is positive when equation (1) is satisfied.

Lemma 1 shows that only nodes that satisfy equation (1) are eligible candidates for reducing total energy consumption. Note that for the source to determine which node is the best relay, the source only needs to know the gain h_rt in addition to any local information the node already has. And, if node s is sending a packet directly to node t, all the eligible candidates can already decode the packet because h_sr>h_st.

Adding the Second Relay Between the Source and the Destination

Let node r denote the optimal first relay already present in the DEA route as shown in FIG. 2A. As shown in FIGS. 2B-2D, the second relay q can be added to one of the three links: s-t, s-r, and r-t. Lemma 2 states that the first possibility is always sub-optimal and need not be considered.

Lemma 2

If the relay r is the optimal single relay for cooperating in the transmission from nodes s to t, adding an additional node, q, in parallel between nodes s and t, as in FIG. 2B cannot reduce the total transmission power in DEA.

Proof

In order for both relays q and r to successfully decode the packet from node s, node s must transmit with a minimum power P_s= γ/max {h_sq, h_sr)}. After nodes q and r successfully decode the packet, it is optimal to add power only to the node with the best channel to node t. Thus, two relays in parallel is only useful if h_qt=h_rt.

Now, assume that h_qt=h_rt. If h_sq>h_sr, then this implies that P_s^red(q)>P_s^red(r), which contradicts the assumption that relay r is the optimal single relay. If h_sq<h_sr, then only node r should be used as the relay. If h_sq=h_sr, then the total power consumption is the same as the single relay case.

Based on Lemma 2, we only need to consider adding a new relay between the s-r and r-t links in the established DEA route, as shown in FIGS. 1C-1D.

Lemma 3

Let node r be the optimal single relay in an established DEA route. If and only if there exists a node q ε V−{s, r, t}, such that h_sq>h_sr, h_qt<min{h_qr, h_rt}, and

h _qr((1/h_sr)−(1/h_sq))>(h_rt−h_qt)/(h_rt−h_st),  (5)

then adding node q between nodes s and r, as in FIG. 1C, reduces total energy consumption. The optimal power consumption, P_s^red(q), is

P _s ^red(q)= γ/h_rt[(h_rt−h_st)((1/h_sr)−(1/h_sq))+((h_qt−h_rt)/hqr),  (6)

• • when the source and the relays set their respective transmission powers P_s, P_q, and P_r, at

P _s=1/h_sqγ, P_q=1/h_qrγ, and P_r=(1/h_rt)(1−h_st/h_sq−h_qt/h_qr) γ  (7)

Proof

In an energy efficient DEA route, each relay transmits the packet with the minimum power required to reach the next relay, while the last relay transmits the packet to the destination with a power that is just sufficient for the destination to decode the packet using the energy accumulated from the transmissions by previous relays. This can be shown to lead to the power allocation in equation (7) for the DEA route s-q-r-t. The power reduction in equation (6) is the difference between the total transmit powers for routes s-q-r-t and s-r-t.

The DEA route s-q-r-t cannot reduce power if h_sq>h_sr, otherwise, node q can be dropped from the route, as node r itself can successfully decode the packet transmitted by node s. Similarly, node r can be dropped from the route if h_qt>min{h_qr, h_rt}. But this contradicts the assumption that node r is the optimal single relay. The total power reduction in equation (6) is positive if and only if the condition in equation (5) is satisfied.

Lemma 4

Let node r be the optimal single relay in an established DEA route. If and only if there exists a node qεV−{s, r, t}, such that

h _qt >h _rt, and h_rt/h_rq<1−h_st/h_sr,  (8)

then adding node q between nodes r and t, as shown in FIG. 1D, leads to an optimal power reduction, P_r^red(q), of

P _r ^red(q)=(1/h_rt−1/h_qt)(1−h_sth_sr−h_rt/h_rq) γ,  (9)

when the source and the relays set their transmission powers P_s, P_q, and P_r, respectively, at

P _s=1/h_srγ, P_r=1/h_rqγ, and P_q=1/h_qt(1−h_st/h_sr−h_rt/h_rq) γ.  (10)

Proof

The power allocation in equation (10) follows from an argument similar to that in Lemma 3. Also, node q can be dropped from the DEA route s-r-q-t if h_qt≦h_rt. The total power reduction in equation (9) is the difference between the total powers consumed by routes s-r-q-t and s-r-t. It is positive if and only if equation (8) is satisfied.

Notice that before the second relay is added, the first relay r transmits the packet with power 1/h_rt(1−h_st/h_sr) γ. From the necessary and sufficient condition in equation (8), it can be seen that all eligible nodes that can reduce total power can successfully decode the packet transmitted by node r. This fact is exploited when we progressively add relays to reduce the total power consumption.

Multiple Relays

As described above, two relays in parallel cannot reduce the total power consumption over an optimal single relay DEA route. This result can be generalized to the case where multiple relays are present. Therefore, we only need to consider the cases where new nodes are inserted in between two adjacent relays or between a relay and the destination, as was done in FIGS. 1C-1D. We refer to such a route a serial DEA route.

To consider adding a node, w, in the serial DEA route that already contains multiple relays, we first define the following terminology. If nodes u and v are two relays in the serial DEA route, and node u successfully decodes the packet before the relay v, then we say that node u is before node v, and node v is after node u. We say that node v is immediately after or next to node u if node v is after node u, and there is no relay that is after node u and before node v. The relay immediately after node u in the serial DEA route is denoted by N(u). A relay u is called the last relay in the serial DEA route if N(u)=t.

The relay set, R, is the set of all relays, excluding the destination, that are in the serial DEA route. The backward relay set, B(u), is the ordered set of relays before node u in the route. A(u)=P_rεB(u)h_rt/h_rN(r) denotes the fraction of the total energy, which is required to successfully decode a packet at the destination. The energy accumulates at the destination due to transmissions from the relays in the set B(u).

Theorem 1

Let u be a relay in the serial DEA route, with v=N(u) being the relay immediately after the relay u. If u is not the last relay, 1, in the route, then adding the node was a relay immediately after node u reduces the total power consumption if w satisfies the following two sufficient conditions:

h _uw >h _uv and h _wv(1/h_uv−1/h_uw)>(h_lt−h_wt)/(h_lt−h_ut).  (11)

A total power reduction of

P _u ^red(w)=1/h_lt[(h_lt−h_ut)(1/h_uv−1/h_uw)+(h_wt−h_lt)/h_wv)] γ  (12)

is achieved when the transmit powers node of u and l are changed to

P _u = γ/h _uv and P_l=1/h_lt(1−A(l)+h_ut/h_uw−h_wt/hwy) γ.  (13)

The transmit power of the new relay, w, is P_w= γ/h_wv. The transmit powers of all the other relays in the route are unchanged.

Proof

Using an argument analogous to that in Lemma 3, the power allocation after node w is added as a relay corresponds to that in equation (13). The condition for power reduction in equation (11) can be derived in a fashion similar to equation (5). To achieve the power reductions, the condition in equation (11) requires that every relay in the serial DEA route determines the gain h_lt. This is not conducive to a distributed implementation. The following corollary provides a sufficient condition that guarantees power reductions without the need for every relay determining the gain h_lt.

Corollary 1

When node u is not the last relay in the serial DEA route, adding the node w immediately after node u results in power reductions if

h _wt >h _ut and 1/h_uw+1/h_wv<1/h_uv.  (14)

Theorem 2

When node u is the last relay in a serial DEA route, adding a node w immediately after node u can reduce power consumption if w satisfies the two conditions:

h _wt >h _ut and h_ut/h_uw<1−A(u).  (15)

A total power reduction of

P _u ^red(w)=(1/h^ut−1/h_wt)(1−A(u)−h_ut/h_uw) γ  (16)

is achieved when the transmit power of node u is changed to P_u= γ/h_uw, and the transmit power of the new node w is

P _w=1/h_wt(1−A(u)−h_ut/h_uw)  (17)

The transmit powers of all the other relays in the route are unchanged.

Proof

Using an analogous argument as in Lemma 4, the power allocation, after node w is added, corresponds to that in equation (10). The condition for power reduction in equation (15) can be derived in a similar manner as in equation (8). Both theorem 2 and corollary 1 show that all potential relays, i.e., the nodes that lead to power reductions, can already successfully decode the transmissions from immediately previous relays. As a result, local channel state information and minimal feedback from the potential relays can be used to progressively increment the serial DEA route to reduce total power.

Progressive Accumulative Route (PAR)

Initially, a basic route is established between the source and the destination. Conventional route discovery processes can be used to discover a route between nodes s and t in networks when a direct link from node s to t does not exist.

Then, the PAR process progressively and distributively adds relays to improve the energy-efficiency of the serial DEA route. That is, energy consumption is reduced while transmitting packets along the route. This relay discovery process is done via two types of packets: a data packet that contains the data to be transmitted from node s to node t, and a ready-to-cooperate (RTC) packet for feedback of the limited additional information required for modifying the route.

The source transmits data to the destination through the already established serial DEA route. The source transmits a new packet to its next relay, N(s), with power γ/h_sN(s). Neighboring nodes that receive a transmission from a currently transmitting relay in the established serial DEA route check, using only the local information available and the information in the data packet, whether their participation as a relay can lead to further power reduction. If so, the nodes feedback the RTC packet to the relay whose transmission the nodes overheard.

FIG. 3 shows the structure of the data and RTC packets. The meaning of each field in the packets is shown in FIG. 4.

The pseudo code of the PAR process is shown in FIG. 5. When a relay u, which is not the source, successfully decodes the data packet p, the relay acts upon the packet only if p.RDest=u. Then, the relay knows that the final destination is p.MDest, and the total power that has accumulated at the destination after p was transmitted is p.FracDelivered+p.GainD/p.GainR. That is, the relay records the fractional energy that will be accumulated at the destination due to transmitting a particular packet to the destination. This recorded information is also forwarded to other nodes along the route, so that those nodes can also participate in the design of the route that minimizes energy.

If u is not the last relay, it transmits the packet to its next relay with power γ/h_uN(u). If the node is the last relay, the node transmits the packet with power (1−A(u))/h_ut.

The relay u updates the route after a sufficient time, minTime, has elapsed since it last updated the route. The time minTime depends on a multiple access protocol, and is used to ensure that a relay has sufficient time to receive RTC feedback packets before the node decides on an additional relay. The node updates the next relay to be the next node, denoted by bestCandidate. This leads to maximum power reduction. The RTC packets enable node u to find the node bestCandidate. When node u receives the RTC packet from node w, the fields of the packet enable node u to determine the power reduction if node w is made the next relay as follows.

If u is not the last relay, P_u^red(w)=(1/h_uv−1/h_uw−1/h_wv) γ,  (18)

If u is the last relay, P_u^red(w)=(1/h_ut−1/h_wt)(1−A(u)−h_ut/h_uw) γ,  (19)

where v is the relay immediately after u: v=N(u). If P_u^red(W) exceeds the power reduction achievable by the current best candidate, we update bestCandidate to be node w.

When the node w receives the data packet, p, from the relay U, the fields of the data packet enables node w to check, using equations (14) or (15), whether becoming a relay can reduce total power. If so, node w stores N(w)=p.RDest in memory, and generates and transmits an RTC packet to u when possible, according to multiple access protocol. The pseudo code for a node is given in FIG. 6.

EFFECT OF THE INVENTION

In the wireless relay network according to embodiments of the invention, only the destination accumulates energy, but the relay nodes do not. Such network, with considerably simpler relays, has comparable energy efficiency as a conventional network where energy accumulates at every node. A destination energy accumulative network is also more energy efficient than traditional multi-hop networks that do not accumulate energy.

The PAR process discovers the DEA route and determines the relay transmission powers in a distributed manner. The process exploits local information about the channel gains, and uses very limited feedback from nodes that can be added to the route as relays. The route discovery in PAR has a very low complexity, and is in contrast to the NP-complete nature of the route discovery process in full energy accumulative networks.

Using PAR, the nodes receive and can decode the packets currently being transmitted in the DEA route, and determine whether the nodes can act as relays to reduce the total power consumption of the route.

The latency for route setup using PAR is low, because a basic connectivity between the source and the destination is established right from the beginning, and improved routes, which progressively add more relays, over time. PAR is well suited for reducing the energy consumption in practical sensor networks with low complexity nodes.

Although the invention has been described by way of examples of preferred embodiments, it is to be understood that various other adaptations and modifications may be made within the spirit and scope of the invention. Therefore, it is the object of the appended claims to cover all such variations and modifications as come within the true spirit and scope of the invention.

Claims

1. A method for transmitting packets from a source node to a destination node via relay nodes of a wireless network, comprising the steps of: transmitting, from a source node, a packet along a route of relay nodes to a destination node in a wireless network;receiving, in the destination node, multiple versions of the packet;storing, only in the destination node, the multiple versions of the packet, anddecoding, in the destination node, the packet using the multiple received versions of the packet.

2. The method of claim 1, in which the destination node receives the multiple versions of the packet from multiple relay nodes.

3. The method of claim 1, in which the versions of the packet are copies of the packet.

4. The method of claim 1, in which the nodes use at least one antenna.

5. The method of claim 1, further comprising: determining progressively an energy efficient route from the source node to the destination node.

6. The method of claim 5, in which the energy efficient route is based only on local channel state information available at each node.

7. The method of claim 5, further comprising: setting transmit powers in the nodes.

8. The method of claim 7, in which the setting is done in a distributed and progressive manner.

9. The method of claim 1, in which energy of the multiple versions of the packet is accumulated only by the destination node.

10. The method of claim 1, in which the decoding of the packet uses an accumulated energy of the received multiple versions of the packet.

11. The method of claim 10, in which the packet is decoded after the accumulated energy of the multiple versions of the packet is equal to or greater than a predetermined threshold.

12. The method of claim 11, in which the predetermined threshold depends on a modulation and a coding used for transmitting the packet.

13. The method of claim 1, in which a cyclic redundancy check is inserted in the packet to determine whether the packet is decoded correctly.

14. The method of claim 1, further comprising: adding a relay node to the route so as to reduce a total power consumption along the route.

15. The method of claim 1, in which the added relay node is inserted serially between nodes along the route.

16. The method of claim 1, in which the packet is received by a potential relay node.

17. The method of claim 16, in which a request-to-cooperate packet is sent by the potential relay node to indicate that the potential relay node can act as a relay node along the route.

18. The method of claim 16, in which a particular node receiving the request-to-cooperate packet determines whether to use the potential relay node.

19. The method of claim 1, further comprising: determining the route from the source node to the destination node; andadding progressively relay nodes to the route to decrease energy consumption while transmitting the packet along the route.

20. The method of claim 1, further comprising: updating progressively the route from the source node to the destination node to decrease the energy consumption.

21. The method of claim 1, in which a particular relay node records a fraction of energy accumulated at the destination node due to transmitting a particular packet to the destination node.

22. The method of claim 21, further comprising: forwarding the recorded fraction of energy to a next node along the route.

23. The method of claim 1, in which a particular relay node records a fraction of energy accumulated at the destination node due to transmitting a particular packet to the destination node by nodes along the route from the particular relay node to the destination node.

24. A system for transmitting packets from a source node to a destination node via relay nodes of a wireless network, system comprising: a source node configured to transmit a packet;a plurality of relay nodes configured to only receive the packet and only to retransmit the packet; anda destination node configured to receive multiple versions of the packet, and only the destination node configured to store the received multiple version of the packet, and the destination node including means for decoding the packet using the multiple received versions of the packet.