Refined Routing Algorithm in Hybrid Networks with Different Transmission Rates Won-Yong Shin, Member, IEEE

Abstract—In this paper, a low-complexity routing strategy is introduced for a hybrid network in which 𝑛 wireless nodes are randomly located and multiple base stations (BSs) are used. We allow different transmission rates according to the geographic location of each source in the network. Our achievable scheme consists of the following two routing schemes: a BS-based singlehop (SH) routing in the up/downlink and conventional multihop transmission. Specifically, in the proposed BS-based scheme, when a source-destination (S-D) pair is inside the center region with respect to its BS, the source is allowed to transmit data to its closest BS via SH. Our results indicate that a logarithmic boost (i.e., power gain) is provided in the throughput scaling law, compared to the conventional schemes, as the number of BSs is greater than a certain threshold. The gain comes from the fact that S-D pairs in their center regions can exploit pernode transmission rates log 𝑛 due to the increased received signal power from/to their closest BS. Index Terms—Base station (BS), center region, hybrid network, logarithmic boost, multi-hop, power gain, routing, scaling law, single-hop (SH), source-destination (S-D) pair, throughput.



N their seminar work, Gupta and Kumar [1] introduced and characterized the sum-rate scaling in a large wireless ad hoc network. They showed that, for a network of 𝑛 sourcedestination (S-D) pairs randomly √distributed in a unit area, the total throughput scales as Θ( 𝑛/ log 𝑛).1 This throughput scaling is achieved using a multi-hop (MH) communication scheme. A recent result has been shown that an almost linear throughput in the network, i.e., Θ(𝑛1−𝜖 ) for an arbitrarily small 𝜖 > 0, is achievable by using a hierarchical cooperation strategy [3], which is the best we can hope for. Besides the schemes in [3], there has been a steady push to improve the throughput of wireless networks up to a linear scaling in a variety of network scenarios—a form of multi-user diversity is exploited in networks with mode mobility [4], and interference alignment schemes [5] are used in interferencelimited networks. Since to achieve such a linear scaling in wireless networks there will be a price to pay in terms of higher delay and higher cost of channel estimation, it would also be good

Paper approved by T. T. Lee, the Editor for Switching Architecture Performance of the IEEE Communications Society. Manuscript received February 19, 2010; revised June 4, 2010, August 10, 2010, and September 30, 2010. W.-Y. Shin is with the School of Engineering and Applied Sciences, Harvard University, Cambridge, MA 02138 USA (e-mail: [email protected]). Digital Object Identifier 10.1109/TCOMM.2011.020411.100041 1 We use the following notations: i) 𝑓 (𝑥) = 𝑂(𝑔(𝑥)) means that there exist constants 𝐶 and 𝑐 such that 𝑓 (𝑥) ≤ 𝐶𝑔(𝑥) for all 𝑥 > 𝑐. ii) 𝑓 (𝑥) = 𝑜(𝑔(𝑥)) (𝑥) means that lim 𝑓𝑔(𝑥) = 0. iii) 𝑓 (𝑥) = Ω(𝑔(𝑥)) if 𝑔(𝑥) = 𝑂(𝑓 (𝑥)). 𝑥→∞ iv) 𝑓 (𝑥) = 𝜔(𝑔(𝑥)) if 𝑔(𝑥) = 𝑜(𝑓 (𝑥)). v) 𝑓 (𝑥) = Θ(𝑔(𝑥)) if 𝑓 (𝑥) = 𝑂(𝑔(𝑥)) and 𝑔(𝑥) = 𝑂(𝑓 (𝑥)) [2].

to have infrastructure aiding wireless nodes. Such hybrid networks consisting of both wireless ad hoc nodes and infrastructure nodes, or equivalently base stations (BSs), have been introduced and analyzed in [6]–[9]. BSs are assumed to be interconnected by high-capacity wired links. It is strictly necessary for the number 𝑚 of BSs to exceed a threshold in order to obtain a linear throughput scaling in 𝑚. In [10], optimal capacity scaling was characterized for a more general hybrid network, where multiple antennas are deployed at each BS and the achievability result is based on using one of BSbased single-hop (SH) and MH routings, conventional MH transmission [1], and hierarchical cooperation strategy [3]. To guarantee such a scaling result, however, it is crucial to assume channel state information (CSI) at each BS and multi-user detection schemes. In this paper, we introduce a novel low-complexity routing strategy using 𝑚 regularly-placed BSs, each of which has a single antenna. We then show a theoretical evaluation of the proposed algorithm in terms of throughput scaling, while providing a better throughput with easier implementation than that of the conventional schemes [6]–[9]. In practice, higher data rates can be exploited by users who are closer to the BS in their cell for cellular environments, which is rather obvious. We take into account the scenario where different transmission rates are allowed, depending on the geographic location of each source in the network. In this case, the goal of this work is to exploit both degrees-of-freedom (DoF) and power gains in hybrid networks, whereas since the benefits of infrastructure support are not fully understood, a DoF gain is only obtained by using the conventional BS-supported routings [6]–[9]. To be specific, our achievable scheme is composed of using one of the following two routing algorighms: a direct SH routing to each BS in the uplink and from each BS in the downlink and conventional MH transmission [1], [11] without help of BSs. In the proposed BS-based scheme, transmission via SH routing is allowed only when both a source and its destination are on their circles with a certain (small) radius, called center regions, where each BS is located at the center of its circle. It is shown that there exists a logarithmic boost in the throughput scaling law, compared to the conventional schemes [6]–[9], as 𝑚 is greater than a certain value. This is because per-node transmission rates log 𝑛 can be achieved by each S-D pair in the center region due to the increased signal power from/to the closest BS. That is, the proposed BS-based scheme leads us to obtain a power gain from the close range transmission from/to BSs. The rest of this paper is organized as follows. Section II describes our system and channel models. In Section III, our routing algorithm is described. In Section IV, the performance

c 2011 IEEE 0090-6778/11$25.00 ⃝



where 𝑥𝑖 ∈ 𝒞 is the signal transmitted by the 𝑖-th node, 𝑛𝑘 denotes the circularly symmetric complex additive white Gaussian noise with zero-mean and variance 𝑁0 . The channel gain ℎ𝑘𝑖 between node 𝑖 ∈ {1, ⋅ ⋅ ⋅ , 𝑛} and BS 𝑘 is given by ℎ𝑘𝑖 =

Fig. 1.

The wireless ad hoc network with infrastructure support.

of the proposed scheme is analyzed. Finally, Section V summarizes the paper with some concluding remarks. II. S YSTEM AND C HANNEL M ODELS We consider a two-dimensional wireless network that consists of 𝑛 S-D pairs uniformly and independently distributed on a square of unit area in dense networks [1], [3], [11].2 Suppose that the whole area is divided into 𝑚 square cells, each of which is covered by one single-antenna BS at its center (see Fig. 1), which similarly follows those in [6]–[9].3 No nodes are assumed to be physically located inside the BSs. Parameters 𝑛 and 𝑚 are related according to 𝑚 = 𝑛𝛽 , where 𝛽 ∈ (0, 1). Furthermore, it is assumed that the BSto-BS links have infinite capacity and that these BSs are neither sources nor destinations. Suppose that each node and BS has an average transmit power constraint 𝑃 (constant). CSI is assumed to be available at the receivers but not at the transmitters. It is assumed that each node can transmit its data packet at a different rate according to its geographical location in the network. We use 𝑇 (𝑛) to denote the total throughput of the network. We do not assume the use of any sophisticated multi-user detection schemes at each receiver, thereby resulting in easier implementation. The signal model in the uplink is now described as follows. Let 𝐼 ⊂ {1, ⋅ ⋅ ⋅ , 𝑛} denote the set of simultaneously transmitting wireless nodes, which is a subset of 𝑛 transmitters available in the network. Then, the received signal 𝑦𝑘 at BS 𝑘 ∈ {1, ⋅ ⋅ ⋅ , 𝑚} at a given time instance is given by ∑ 𝑦𝑘 = ℎ𝑘𝑖 𝑥𝑖 + 𝑛𝑘 , 𝑖∈𝐼 2 An extended network [3], [12], [13] of unit node density can also be considered as another fundamental network model. In extended networks, it is possible to obtain a logarithmic gain, i.e., power gain, under certain conditions by allowing a total transmit power constraint 𝑛𝑃 (as in [12]) for our BS-based transmission, even if the details are not shown in this paper. Our analysis can then be extended to the general case [14] where the network area varies from 1 to 𝑛, and we believe that a power gain would also be obtained even for the network having arbitrary sizes with 𝑛. 3 As one of other BS configurations, suppose that BSs are arbitrarily placed √ maintaining per-BS distance Θ( 𝑚). This does not cause any change in performance, even if the details are not shown in this paper.

𝑒𝑗𝜃𝑘𝑖 𝛼/2




where 𝑒𝑗𝜃𝑘𝑖 represents the random phase uniformly distributed over [0, 2𝜋] and independent for different 𝑖, 𝑘, and time (transmission symbol), i.e., fast fading is assumed.4 Note that this random phase model is based on a far-field assumption [3], [10], which is valid if the wavelength is sufficiently small. 𝑟𝑘𝑖 and 𝛼 > 2 denote the distance between node 𝑖 and BS 𝑘, the path-loss exponent, respectively. Likewise, the complex channel in the downlink ℎ𝑖𝑘 between BS 𝑘 ∈ {1, ⋅ ⋅ ⋅ , 𝑚} and node 𝑖 ∈ {1, ⋅ ⋅ ⋅ , 𝑛}, and the complex channel between nodes 𝑖 and 𝑘 (𝑖, 𝑘 ∈ {1, ⋅ ⋅ ⋅ , 𝑛}) can be modeled in a similar manner. The superscript 𝒞 denotes the absolute complement of a set. Unless otherwise stated, all logarithms are assumed to be to the base 2. III. P ROPOSED ROUTING A LGORITHM In this section, we show our new routing scheme utilizing infrastructure. In addition, the conventional routing without BSs is summarized. We simply use both MH and SH strategies in the cases using the nodes other than S-D pairs as relays. To avoid a large interference, different time slots are used between routing schemes with and without infrastructure support. A. Infrastructure-Supported Routing In the network using BSs, when a source node’s data passes through its nearest BS, i.e., access routing is performed, via MH routing as in [6]–[9], the throughput per S-D pair is given by a constant scaling. Now we explain a single-cell operation and its fundamental limit. Then, the analysis is related to that for many-to-one channels [17] (in the exit routing, it is converted to that for one-to-many channels). In particular, the transmission scheme and its achievable rates were shown in [17] for an uplink network in which all nodes are distributed uniformly over the boundary of a circle with a unit radius and the BS is at the center of the circle. Here, it was proved that rates log 𝑛 can be achieved by coherent combining techniques, which is order-optimal in the single-cell network model. We remark that such a logarithmic gain (i.e., power gain) does not appear in the conventional BS-supported schemes [6]–[9]. This motivates us to construct a new routing scheme in hybrid networks having multiple BSs, especially with shortrange SH transmission to BSs, thus yielding a power gain as well as a DoF gain. Since there exist both access (to BSs) and exit (from BSs) routings, we also use different time slots, e.g., 4 Under the propagation model, the received signal power from the desired transmitter can be larger than the transmitted power, which is not reasonable. To avoid such a problem, an alternative propagation model has been taken 𝛼/2 into account in [15], [16], where (1 + 𝑟𝑘𝑖 )𝛼/2 is adopted instead of 𝑟𝑘𝑖 . However, the model defined by (1) has been commonly used to analyze throughput scaling laws in ad hoc networks [1], [3], [4], [10] since in dense networks, power is not a matter of interest and interference management is important in determining performance.


Fig. 2.


The infrastructure-supported routing.

even and odd time slots, respectively. The proposed scheme shown in Fig. 2 under dense networks is now proposed as follows: ∙ Divide the network into equal square cells of area 1/𝑚 having one BS at the center of each cell, where each square cell represents the Voronoi tessellation.5 Note that each cell obviously has its center region at the center, which is illustrated in Fig. 2. 1 ∙ Let denote the radius of the center region for 𝑚1/2+𝜖0 some 𝜖0 > 0 independent of 𝑛. Then, we set the parameter 𝜖0 such that each center region includes at least one source whose destination is also located in one of the other center regions (which will be rigorously discussed in the following section). Under the setup, S-D pairs only in their center regions are selected for transmission through BSs. ∙ For the access routing (to BSs), a source node in each center region transmits its packets via SH to the BS in the same cell. A transmit power of 𝑚𝑃𝛼/2 is used at each node, where 𝑃 is per-node transmit power constraint. ∙ The BS that completes decoding its packets transmits them to the BS that is nearest to the corresponding destination via wired BS-to-BS links. ∙ For the exit routing (from BSs), each BS transmits the received packets via SH to the corresponding destination in the center region of its cell. A transmit power of 𝑚𝑃𝛼/2 is used at each BS as in the uplink case. The given scheme satisfies the average power constraint 𝑃 since the transmit power 𝑚𝑃𝛼/2 at each node (or BS) tends to zero as 𝑛 → ∞ for 𝛼 > 2. Note that the scheme does not require any CSI at the transmitters unlike those in [17]. B. Conventional MH Routing To improve throughput scalings of BS-supported networks, the number 𝑚 of BSs should be higher than a certain level. That is, pure ad hoc transmissions without help of BSs may achieve better throughput scaling when 𝑚 is not large enough. In addition, to enable S-D pairs outside the center regions to have their own traffic demands, the conventional MH scheme not using BSs is applied in the network. In this subsection, we briefly introduce how to operate the MH routing proposed 5 If BSs are regularly distributed on a planar disk domain, then hexagontype Voronoi cells are generated similarly to [6], [7], [9]. Since the shape of tessellated Voronoi cells does not essentially change the throughput scaling, for analytical convenience we simply assume the square network as in the case of [8].

in [1], [11]. We first show the following lemma about the geometry of 𝑛 nodes in the network divided into equal routing cells of area 𝑎(𝑛). Lemma 1: If 𝑎(𝑛) ≥ 2 log 𝑛/𝑛, then each routing cell has at least one node with high probability (whp). The proof of this lemma is presented in [1], [11]. The basic procedure of the MH protocol in dense networks is then described as follows: ∙ Divide the network into square routing cells of area 2 log 𝑛/𝑛. ∙ Draw a line connecting an S-D pair. Since each cell has at least one node from Lemma 1, it follows that a source transmits a packet to its destination using one of the nodes in each adjacent routing cell passing through the line. 𝑃 ∙ A transmit power of is enough to guarantee the 𝑛𝛼/2 required throughput scaling of the MH routing at each transmission. ∙ Each routing cell operates the 9-time division multiple access to avoid a large interference. Note that since there are Θ(𝑛/ log 𝑛) routing cells in the network and the number of hops per S-D pair is given by √ √ 𝑂( 𝑛/ log 𝑛), Ω( 𝑛/ log 𝑛) source nodes can be activated simultaneously. IV. P ERFORMANCE A NALYSIS In this section, the throughput scaling of our achievable scheme is analyzed and compared with the best achievable result [6] among the conventional strategies [6]–[9] in the network with BSs. We start from the following lemma. Lemma 2: Suppose 𝑚 = 𝑛𝛽 and 0 < 𝜖0 < 1−𝛽 2𝛽 where 𝛽 ∈ (0, 1). Then, the number of nodes inside each center region is 1−𝛽(1+2𝜖0 ) , (1 + 𝛿0 )𝜋𝑛1−𝛽(1+2𝜖0 ) ), i.e., between ) − 𝛿0 )𝜋𝑛 ( 𝑛 ((1 Θ 𝑚1+2𝜖0 , whp for some constant 0 < 𝛿0 < 1 independent of 𝑛. The proof of this lemma is given by slightly modifying the asymptotic analysis in [3]. Then, we turn our attention to deciding a feasible size, given by a function of 𝜖0 , of the center region. When we use ℰ𝑙 to denote the event that the 𝑙-th center region includes at least one source such that its destination exists in one of the other center regions, we obtain the following result. Lemma 3: If 0 < 𝜖0 < 1−𝛽 4𝛽 , then the event ℰ𝑙 holds whp for each 𝑙 ∈ {1, ⋅ ⋅ ⋅ , 𝑚}, where 𝛽 ∈ (0, 1). Proof: Suppose 0 < 𝜖0 < 1−𝛽 4𝛽 . From Lemma 2, it is then 0𝑛 seen that each center region has 𝑚𝑐1+2𝜖 nodes whp, where 𝑐0 0 is a constant independent of 𝑛. Since nodes are uniformly distributed in the network, the probability that the event ℰ𝑙 does not hold for the 𝑙-th center region is thus given by ( )𝑐0 𝑛/𝑚1+2𝜖0 𝜋 { 𝒞} 𝑚1+2𝜖0 . 𝑃 ℰ𝑙 = 1 − 1 𝑚

By the union bound over all the center regions, we have the following inequality: ( )𝑐0 𝑛/𝑚1+2𝜖0 𝜋 𝑚1+2𝜖0 𝑃 {ℰ1 ∪ ⋅ ⋅ ⋅ ∪ ℰ𝑚 } ≥ 1 − 𝑚 1 − 1 𝑚

1−𝛽−2𝛽𝜖0 ( 𝜋 )𝑐0 𝑛 = 1 − 𝑚 1 − 2𝛽𝜖0 ) (𝑛 = 1 − 𝑚 exp −𝑐0 𝜋𝑛1−𝛽−4𝛽𝜖0 , (2)


Fig. 3. Grouping of interfering center regions. The first layer 𝑙1 of the network represents the outer 8 shaded cells including center regions.

which tends to one as 𝑛 goes to infinity under the assumption 0 < 𝜖0 < 1−𝛽 4𝛽 . This completes the proof of this lemma. Now we are ready to present the achievable total throughput 𝑇 (𝑛) in the network with BSs. Theorem 1: In a dense network using our BS-based routing, ⎧ (√ )  𝑛/ log 𝑛  ⎨ Ω 𝑇 (𝑛) = Ω (𝑚 log 𝑛)   ⎩

) 𝑛/(log 𝑛)3/2 ) (√ 𝑛/(log 𝑛)3/2 if 𝑚 = Ω ( ) and 𝑚 = 𝑂 𝑛1−𝜖

if 𝑚 = 𝑜



is achievable for all 𝑚 = 𝑛𝛽 satisfying 𝛽 ∈ (0, 1), where 𝜖 > 0 is an arbitrarily small constant. Proof: First we consider the uplink case (access routing). There are 8𝑘 interfering center regions, each of which includes one transmitting source, in the 𝑘-th layer 𝑙𝑘 of the network as illustrated in Fig. 3. Let 𝑑𝑘 denote the Euclidean distance between a given BS and any transmitting source in the center regions √ of 𝑙𝑘 , which is a random variable. Since 𝑑𝑘 scales as Θ(𝑘/ 𝑚), there exists 𝑐3 > 𝑐2 > 0 with √ constants 𝑐2 and 𝑐3 independent of 𝑛, such that 𝑑𝑘 = 𝑐1 𝑘/ 𝑚, where all 𝑐1 lies in the interval [𝑐2 , 𝑐3 ]. Then, the signal-to-interference-and-noise ratio (SINR) at each BS is given by ( 1/2+𝜖 )𝛼 𝑃 0 𝑚 𝑚𝛼/2 SINR ≥ 𝑃 ∑ −𝛼 + 𝑁0 𝑖∈𝐼 𝑑𝑖 𝑚𝛼/2 𝜖0 𝛼 ≥ 𝑐4 𝑃 𝑚 , where 𝐼 denotes the set of simultaneously transmitting nodes (in all the center regions) and 𝑐4 is a constant independent of 𝑛. Here, the second inequality holds since ∑ 𝑖∈𝐼

𝑑−𝛼 ≤ 𝑖

∞ ∑ 𝑖=1

∞ 8𝑚𝛼/2 ∑ 1 8𝑖 √ 𝛼 = ≤ 𝑐5 𝑚𝛼/2 𝛼−1 𝑐𝛼 𝑖 (𝑐2 𝑖/ 𝑚) 2 𝑖=1

for some constant 𝑐5 > 0 independent of 𝑛. Hence, by assuming the full CSI at the receiver, the transmission rate at each BS can be lower-bounded by log(1 + SINR) = Ω(log 𝑛) under the conditions 𝛽 ∈ (0, 1) and 0 < 𝜖0 < 1−𝛽 4𝛽 . Similarly, for the exit routing, the rate of Ω(log 𝑛) at each (destination) node can be obtained. Since there are 𝑚 S-D pairs in all the center regions from the result of Lemma 3, the aggregate rates of our BS-based scheme are thus given by 𝑇 (𝑛) = Ω(𝑚 log 𝑛). In addition, when we apply the

Fig. 4.


The performance comparison on the throughput scaling.

conventional √routing [1] that does not utilize the BSs, we get 𝑇 (𝑛) = Ω( 𝑛/ log 𝑛). Based on the two achievability results, it is possible to achieve a lower bound on the total throughput ( { }) √ 𝑛 𝑇 (𝑛) = Ω max 𝑚 log 𝑛, , log 𝑛 which finally yields (3). When the conventional BS-supported scheme in [6] is used in (the network, the total ) throughput is given by 𝑇 (𝑛) = √ Ω max{𝑚, 𝑛/ log 𝑛} , thus resulting in ⎧ (√ ) (√ )  Ω 𝑛/ log 𝑛 if 𝑚 = 𝑜 𝑛/ log 𝑛  ⎨ (√ ) 𝑇 (𝑛) = Ω (𝑚) if 𝑚 = Ω 𝑛/ log 𝑛   ) ( ⎩ and 𝑚 = 𝑂 𝑛1−𝜖 , where 𝜖 > 0 is an arbitrarily small constant. Now in Fig. 4, we verify the benefits of our BS-based routing over conventional achievable schemes above by showing how the total throughput 𝑇 (𝑛) scales with respect to the number 𝑚 of BSs in the network. 𝑅𝐴 and 𝑅𝐵 denotes the proposed and conventional (BS-supported) routings, respectively. We observe that 𝑇 (𝑛) does not increase in both cases as 𝑚 is below a certain level, in which the infrastructure is not helpful. On the other hand, as 𝑚 exceed the level, the BSbased routings become dominant. For example, when 𝑅𝐴 is used in the network, it is examined that 𝑇 √(𝑛) scales linearly with 𝑚 in the operating regimes 𝑚 = Ω( 𝑛/(log 𝑛)3/2 ) and 𝑚 = 𝑂(𝑛1−𝜖 ). In addition, it is clearly seen that 𝑅𝐴 exhibits a much better throughput scaling than√ 𝑅𝐵 while giving us a logarithmic gain on 𝑇 (𝑛) for all 𝑚 ∈ [ 𝑛/ log 𝑛, 𝑛1−𝜖 ]. This gain comes from the fact that a higher received signal power from/to the closest BS enables each S-D pair in the respective center regions to exploit transmission rates log 𝑛. We next use computer simulations to further validate the performance for finite parameters 𝑛 and 𝑚. We slightly modify our system model so that it is suitable for numerical evaluation. To remove any randomness of node location, a regular network is assumed where nodes are equally spaced over a square of unit area. Suppose that there are 1024 regularly-spaced nodes in the whole network and the pathloss exponent 𝛼 is given by 4. We evaluate the total throughput





Total throughput T(n) (bit/s/Hz)




35 30



[1] P. Gupta and P. R. Kumar, “The capacity of wireless networks," IEEE Trans. Inf. Theory, vol. 46, pp. 388-404, Mar. 2000. [2] D. E. Knuth, “Big Omicron and big Omega and big Theta," ACM SIGACT News, vol. 8, pp. 18-24, Apr.-June 1976. [3] A. Özgür, O. Lévêque, and D. N. C. Tse, “Hierarchical cooperation achieves optimal capacity scaling in ad hoc networks," IEEE Trans. Inf. Theory, vol. 53, pp. 3549-3572, Oct. 2007.

20 15 10 5 0 0

pair than the others in ad hoc networks with BSs, assuming non-unform transmission rates. As a result, it was shown that there is always a logarithmic gain further over the conventional achievable schemes in the total √ throughput 𝑇 (𝑛) as the number 𝑚 of BSs is greater than 𝑛/ log 𝑛.


40 60 The number of BSs (m)



Fig. 5. The simulation result for the case where 𝑛 = 1024 and 𝑚 ∈ [4, 100].

𝑇 (𝑛) (bit/s/Hz) according to 𝑚 varying from 4 to 100. For the proposed BS-based SH routing, from the two conditions log 𝑚 𝑚 = 𝑛𝛽 and 0 < 𝜖0 < 1−𝛽 4𝛽 , we set 𝛽 = log 𝑛 and 1−𝛽 𝜖0 = 8𝛽 to specify the center region in a cell. Note that at the finite 𝑛 regime, we need to take into account the probability 𝑃 {ℰ1 ∪ ⋅ ⋅ ⋅ ∪ ℰ𝑚 } in (2) for computing the total throughput 𝑇 (𝑛). When the proposed scheme is used in the network, the average sum-rates via SH are thus lower-bounded by 𝑚𝑃 {ℰ1 ∪ ⋅ ⋅ ⋅ ∪ ℰ𝑚 } log(1 + SINR). As illustrated in Fig. 5, it is also shown that our BS-based routing outperforms the conventional scheme over all finite 𝑚. Although these curves based on finite parameters 𝑛 and 𝑚 look slightly different from our analytical ones, it can be seen that the trends are similar. V. C ONCLUSION We developed a simple infrastructure-supported routing based on geographic information of wireless nodes while exhibiting a much better throughput scaling compared to the conventional BS-supported schemes. The proposed BS-based scheme does not require any CSI at the transmitters and any multi-user detection at the receivers, and hence, can be easily implemented in practice. We proved that nodes closer to their BS can actually achieve higher throughput per S-D

[4] M. Grossglauser and D. N. C. Tse, “Mobility increases the capacity of ad hoc wireless networks," IEEE/ACM Trans. Networking, vol. 10, pp. 477-486, Aug. 2002. [5] V. R. Cadambe and S. A. Jafar, “Interference alignment and degrees of freedom of the 𝐾 user interference channel," IEEE Trans. Inf. Theory, vol. 54, pp. 3425-3441, Aug. 2008. [6] B. Liu, Z. Liu, and D. Towsley, “On the capacity of hybrid wireless networks," in Proc. IEEE INFOCOM, Mar./Apr. 2003, pp. 1543-1552. [7] U. C. Kozat and L. Tassiulas, “Throughput capacity of random ad hoc networks with infrastructure support," in Proc. ACM MobiCom, Sep. 2003, pp. 55-65. [8] S. R. Kulkarni and P. Viswanath, “Throughput scaling for heterogeneous networks," in Proc. IEEE Int. Symp. Inf. Theory, June/July 2003, p. 452. [9] A. Zemlianov and G. de Veciana, “Capacity of ad hoc wireless networks with infrastructure support," IEEE J. Sel. Areas Commun., vol. 23, pp. 657-667, Mar. 2005. [10] W.-Y. Shin, S.-W. Jeon, N. Devroye, M. H. Vu, S.-Y. Chung, Y. H. Lee, and V. Tarokh, “Improved capacity scaling in wireless networks with infrastructure," IEEE Trans. Inf. Theory, under revision for possible publication. Available: http://arxiv.org/abs/0811.0726. [11] A. El Gamal, J. Mammen, B. Prabhakar, and D. Shah, “Optimal throughput-delay scaling in wireless networks-part I: the fluid model," IEEE Trans. Inf. Theory, vol. 52, pp. 2568-2592, June 2006. [12] A. Jovicic, P. Viswanath, and S. R. Kulkarni, “Upper bounds to transport capacity of wireless networks," IEEE Trans. Inf. Theory, vol. 50, pp. 2555-2565, Nov. 2004. [13] M. Franceschetti, O. Dousse, D. N. C. Tse, and P. Thiran, “Closing the gap in the capacity of wireless networks via percolation theory," IEEE Trans. Inf. Theory, vol. 53, pp. 1009-1018, Mar. 2007. [14] A. Özgür, R. Johari, D. N. C. Tse, and O. Lévêque, “Informationtheoretic operating regimes of large wireless networks," IEEE Trans. Inf. Theory, vol. 56, pp. 427-437, Jan. 2010. [15] O. Arpacioglu and Z. J. Haas, “On the scalability and capacity of singleuser-detection based wireless networks with isotropic antennas," IEEE Trans. Wireless Commun., vol. 6, pp. 8-15, Jan. 2007. [16] E. Duarte-Melo, A. Josan, M. Liu, D. L. Neuoff, and S. S. Pradhan, “The effect of node density and propagation model on throughput scaling of wireless networks," in Proc. IEEE Int. Symp. Inf. Theory, July 2006, pp. 1693-1697. [17] H. El Gamal, “On the scaling laws of dense wireless sensor networks: the data gathering channel," IEEE Trans. Inf. Theory, vol. 51, pp. 12291234, Mar. 2005.

Refined Routing Algorithm in Hybrid Networks with ... - IEEE Xplore

Abstract—In this paper, a low-complexity routing strategy is introduced for a hybrid network in which wireless nodes are randomly located and multiple base ...

588KB Sizes 2 Downloads 352 Views

Recommend Documents

Elastic Routing in Wireless Networks With Directional ... - IEEE Xplore
Abstract— Throughput scaling law of a large wireless network equipping directional antennas at each node is analyzed based on the information-theoretic ...

GreenInfra: Capacity of Large-Scale Hybrid Networks ... - IEEE Xplore
Abstract—The cost-effective impact and fundamental limits of infrastructure support with rate-limited wired backhaul links. (i.e., GreenInfra support), directly ...

Investigating Sensor Networks with Concurrent ... - IEEE Xplore
The background behind this demonstration is described as an one-page poster submission. The goal is to show a flow of tools for quick sensor network modeling, from an high level abstraction down to a system validation, including random network genera

The Viterbi Algorithm - IEEE Xplore
HE VITERBI algorithm (VA) was proposed in 1967 as a method of decoding convolutional codes. Since that time, it has been recognized as an attractive solu-.

An Anycast Routing Strategy with Time Constraint in ... - IEEE Xplore
Tuan Le, Mario Gerla. Dept. of Computer Science, UCLA. Los Angeles, USA. {tuanle, gerla}@cs.ucla.edu. Abstract—Delay Tolerant Networks (DTNs) are sparse ...

Increasing Device Lifetime in Backbone Networks with ... - IEEE Xplore
is the application of sleep modes to network devices. Sleep mode is a low power state, which typically lasts for minutes or hours, during which the device does ...

Cognitive Routing with the ETX Metric - IEEE Xplore
Department of Computer Science. University of California, Los Angeles. Los Angeles, USA. {tuanle, rrabsatt, gerla}@cs.ucla.edu. Abstract—The emergence of ...

DEARER: A Distance-and-Energy-Aware Routing with ... - IEEE Xplore
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information. This article has been accepted for publication in a future ...

Secure Mobile Ad hoc Routing - IEEE Xplore
In mobile ad hoc networks (MANETs), multi-hop mes- sage relay is the common way for nodes to communicate and participate in network operations, making ...

Intra-vehicular Wireless Networks - IEEE Xplore
cost and labor in harness manufacturing and installation, increased design complexity, etc. Wireless communications provides an intriguing alternative to wiring.

Capacity of 3D Erasure Networks - IEEE Xplore
Jul 12, 2016 - Abstract—In this paper, we introduce a large-scale 3D erasure network, where n wireless nodes are randomly distributed in a cuboid of nλ × nμ ...

Sensor Data Cryptography in Wireless Sensor Networks - IEEE Xplore
IEEE TRANSACTIONS ON INFORMATION FORENSICS AND SECURITY, VOL. 3, NO. 2, JUNE 2008. 273. Sensor Data Cryptography in. Wireless Sensor ...

cooperative target tracking in vehicular sensor networks - IEEE Xplore
This subject has been studied in fields such as airborne traffic, computer vision, and wireless sensor networks. A VANET brings out new challenges that should ...

Statistical Multiplexing-Based Hybrid FH-OFDMA ... - IEEE Xplore
of orthogonal frequency-division multiplexing (OFDM)-based ultra-wideband (UWB) indoor radio access networks (RANs). The downlink user capacity is here ...

Distributed Average Consensus With Dithered ... - IEEE Xplore
computation of averages of the node data over networks with band- width/power constraints or large volumes of data. Distributed averaging algorithms fail to ...

IEEE Photonics Technology - IEEE Xplore
Abstract—Due to the high beam divergence of standard laser diodes (LDs), these are not suitable for wavelength-selective feed- back without extra optical ...

An Efficient Geometric Algorithm to Compute Time ... - IEEE Xplore
An Efficient Geometric Algorithm to Compute Time-optimal trajectories for a Car-like Robot. Huifang Wang, Yangzhou Chen and Philippe Sou`eres.

Efficient Pattern Matching Algorithm for Memory ... - IEEE Xplore
intrusion detection system must have a memory-efficient pat- tern-matching algorithm and hardware design. In this paper, we propose a memory-efficient ...

Batch scheduling algorithm for SUCCESS WDM-PON - IEEE Xplore
frames arrived at OLT during a batch period are stored in Virtual. Output Queues (VOQs) and scheduled at the end of the batch period. Through simulation with ...

Discrete Binary Cat Swarm Optimization Algorithm - IEEE Xplore
K. N. Toosi university of Tech. ... its best personal experience and the best experience of the .... the cat and the best position found by members of cat swarm.

A New Algorithm for Finding Numerical Solutions of ... - IEEE Xplore
optimal control problem is the viscosity solution of its associated Hamilton-Jacobi-Bellman equation. An example that the closed form solutions of optimal ...

A DTN Routing and Buffer Management Strategy for ... - IEEE Xplore
Dept. of Computer Science, UCLA. Los Angeles, USA. {tuanle, kalantarian, gerla}@cs.ucla.edu. Abstract—Delay Tolerant Networks (DTNs) are sparse mobile.

A Two-Level Multicast Routing Strategy for Delay ... - IEEE Xplore
Dept. of Computer Science, UCLA. Los Angeles, USA. {tuanle, kalantarian, gerla}@cs.ucla.edu. Abstract—Delay Tolerant Networks (DTNs) are sparse mobile.