Overview:

Over the last few years, considerable research has been done in the area of smart or adaptive-array antennas. Antenna arrays can combat multi-path fading of the desired signal and suppress interfering signals, thereby increasing both performance and capacity of wireless systems. The major digital wireless systems used today can provide substantial additional improvement by deploying antenna arrays along with spatial processing. The main gains achieved when using antenna arrays are range increase, multi-path diversity, interference suppression, capacity increase and data rate increase.

The broad goal of this project is to investigate various approaches to leverage the advantages of antenna array technology at the higher layers of the protocol stack in wireless networks. In particular, the focus of the project is a special class of networks called ad-hoc networks. Some of the plausible gains at the higher layers include more intelligent topology control schemes, more efficient MAC protocols, and better routing protocols. Since smart antenna technology is purely a physical layer technology, existing approaches responsible for each of the aforementioned protocols will indeed work when deployed over smart antenna networks. However, they will not be able to leverage, to the fullest extent, the advantages made possible by the flexible physical layer technology. For example, [2] provides some insight into the inefficient performance of the IEEE 802.11 MAC protocol when deployed over a network equipped with simple directional antennas.

The goal of this project is to design higher layer protocols that completely leverage the capabilities of the under-lying smart antenna technology to improve network performance. Specifically, the objective is to design MAC and routing protocols for ad-hoc networks with different kinds of smart antenna technologies. The project can thus be broadly classified into the two phases of (i) MAC protocol design and (ii) routing protocol design.

The MAC protocol design phase initially began with the design of a MAC protocol for ad-hoc networks with switched beam antennas referred to as the Switched Beam Medium Access (SBMA) protocol, following which a Stream-Controlled Medium Access (SCMA) protocol was designed for ad-hoc networks with MIMO links. Using the insights gained from the design of these specific MAC protocols, a unified MAC layer framework (UmodS) for solving the problem of medium access control in ad-hoc networks with smart antennas in general was proposed. The specific solutions for the different antenna technologies derived from the unified framework, were used to evaluate the performance of the different antenna technologies under varied networks conditions and draw insights into the optimal strategy and antenna technology of operation for the various network conditions.  

The routing protocol design began with the design of a routing protocol for ad-hoc networks with MIMO links that exploits the spatial multiplexing and diversity capabilities of MIMO links in routing protocol operations to improve network performance. MIMO links being the most sophisticated of the antenna technologies, the insights gained from its design are currently being explored to propose a unified framework for the design of routing protocols in ad-hoc networks with any smart antenna technology. Though the focus is with respect to reactive routing protocols, the specific protocol components and ideas are easily extendible to proactive routing protocols as well.

The details of the specific projects are summarized below. The results pertaining to each of the projects can be found in the corresponding publications listed below the summary of the projects.

Results / Status:

(1) SBMA: MAC for Switched beam antennas

This protocol addresses the problem of unnecessary back-offs and potential under-utilization caused due to "deafness" [6] that spurs from the directional transmissions of the control packets. Every node makes use of destination-specific dnav in addition to the conventional nav coprresponding to each beam. While the conventional nav of a beam is set on overhearing an RTS or CTS for an upcoming data transmission in its direction, a dnav is set only when the upcoming transmission is not meant to be in its direction. The problem of unnecesary back-offs arises due to the fact that very few neighbors of a node are informed of its upcoming transmission when directional RTS and CTS are used. This protocol tries to solve this problem by transmitting the control messages in all the directions permissible at a node.

Essentially, RTS messages are transmitted in an opportunistically omni-directional fashion, i.e. the messages are transmitted along all the beams that have not been blocked because of other ongoing transmissions (beams that are not experiencing a busy channel). RTS messages transmitted over beams other than the primary beams (in which the data transfer will occur) are explicitly marked as opportunistic messages. CTS messages are transmitted only in the direction towards the receiver since the collisions caused by transmitting them in other directions outweighs the advantages gained from them. This gives the protocol the name, opportunistic RTS (OP-RTS)/directional (OP-CTS) protocol. When nodes hear regular RTS messages, they update the nav for only the beam along which they overheard the messages. However, when nodes hear opportunistic RTS messages along a beam, they update the dnav corresponding to the source of the concerned message. A node can continue to use a beam for transmission as long as the corresponding nav is idle and the dnav corresponding to its destination is not set.

We also make use of directional information to reduce collisions at receivers. The directional information can be obtained using GPS (global positioning system) or through simple coordination between the nodes. Initially we assume the availability of GPS information. We also have a simple coordination mechanism between the nodes that helps them obtain the directional information. The OP-RTS messages provide more information to the neighbors of the transmitter, to prevent them from causing any collision at the receiver. Specifically, OP-RTS messages are piggybacked with the location information of the receiver of the proposed data transmission. Nodes that overhear the OP-RTS messages, use this additional information to set the conventional nav for the beam corresponding to the direction along which the receiver lies, in addition to setting the node specific dnav for the transmitter. The nav for the beam corresponding to the direction of the receiver is set only if the receiver is within its carrier-sensing distance and this beam is aligned with the already receiving beam of the receiver. This would ensure that the overhearing node does not interfere with either the transmitter or the receiver of the ongoing transmission.

(2) SCMA: MAC for MIMO links:

The key optimization considerations and mechanisms in SCMA are summarized below:

1. Stream Control Gains: Multiple interfering links operating simultaneously using stream control achieve better overall throughput performance when compared to a scenario in which they operate using TDMA and `k' streams each.
2. Partial Interference Suppression: The flexible interference suppression capabilities of multiple elements arrays helps create additional resources at a node that can be used in additional transmissions (receptions) to provide additional gain.
3. Receiver Overloading: Perfect stream control will not allow a receiver to be overloaded. This could lead to potential under-utilization of the channel resources. Hence it is not always advisable to perform perfect stream control so as to avoid a degradation in peformance.

The following are the key design mechanisms in the design of the distributed MAC protocol for ad-hoc networks employing fully adaptive array antennas with MIMO links.

1. Range Extension: The presence of multiple elements are used to provide spatial multiplexing gain for the transfer of DATA and ACk packets, while they are used to provide range extension for RTS and CTS packets. This has the additional benefit of alleviating collisions in the network.
2. Coloring: Stream control must be performed only on links that belong to multiple contention regions to allow receiver overloading. Hence this necessitates a mechanism that is capable of distinguishing between the two classes of links.
3. Adaptation: Links that belong to a single contention region perform stream control. Hence this requires that such links determine the fair share of resources to be used by each of them in a purely local manner.
4. Coordinated Scheduling: Further since the mechanisms are executed by each node independently, tobe able to ensure that stream control is performed by links belonging to a single (same) contention region, it is crucial that all such links transmit at the same time with their fair share of resources. This calls for a mechanism of co-ordinated scheduling of such links.

(3) UmodS: A Unified MAC Layer Framework for Ad-hoc Networks with Smart Antennas

Despite the specific differences in the characteristics of the antenna technologies, the fact that they still belong to the general umbrella of "smart antenna" technologies leads to an interesting question: Can the different smart antenna technologies be represented in any unified form? Or consequently, Can unified algorithmic frameworks be developed for ad-hoc networks with smart antennas in general? With this question as the basis, in this prpject we explore the problem of a unified approach to medium access control (MAC) in ad-hoc networks with smart antennas. Such an endeavor has the following benefits. First, a unified representation of the physical layer capabilities of the different types of smart antennas can help researchers see the relative merits of the technologies from the perspective of higher layer protocol design. Second, a unified problem formulation, and subsequent derivation of unified algorithms will enable specific aspects of the solutions developed for one class of antennas to be re-used for other classes as long as there are similar sub-problems. Finally, a unified MAC framework for the different classes of smart antennas will provide a very good platform for studying their relative performance trade-offs for varying network conditions.

In this context, we make the following contributions toward developing a unified MAC layer framework for ad-hoc networks with different types of smart antennas including omni-directional antennas, switched-beam antennas, adaptive array antennas, and MIMO links: (i) For the different antenna technologies, we identify the physical layer capabilities, their relevance to the MAC layer design, and MAC layer design considerations specific to the physical layer capabilities, and capture them through a unified representation; (ii) We provide a unified formulation of the problem of medium access control in ad-hoc networks with smart antennas, and show how the unified problem formulation can be applied to derive the specific formulation for a given technology; (iii) We derive unified centralized algorithms based on the above problem formulation; (iv) Finally, using the proposed algorithms we investigate the relative performance trade-offs of the different technologies, and identify key insights into how the technologies compare under varying conditions.

(4) Routing in Ad-hoc Networks with MIMO Links

The focus of this work is to explore the various capabilities of MIMO links but from the perspective of routing layer protocols. We identify whether and how each of the capabilities can translate to improved performance at the routing layer. More specifically, we make the following contributions:

• We identify the capabilities of MIMO links and capture their relevance to routing layer protocols.
• We analyze both theoretically and practically the relative tradeoffs of exploiting the different capabilities of MIMO links.
• We propose a reactive routing protocol whose components are built on the insights gained from the analysis, and hence leverage the PHY layer characteristics in their operations to improve network performance.

Briefly, we identify two fundamental capabilities of MIMO links, namely spatial multiplexing and diversity that can be exploited by the routing layer protocols in their operations.  However, since these two capabilities cannot be fully leveraged at the same time, there exists a trade-off in exploiting these capabilities. It thus becomes necessary to investigate the relative trade-offs between spatial multiplexing and diversity, in order to determine the optimal strategy of operation from the perspective of improving the aggregate network throughput. To this end, we analytically study the benefits and drawbacks of both the strategies from the perspective of routing layer protocols. The study also incorporates practical considerations in determining the optimal strategy of operation. A routing protocol is then proposed with components built on the insights gained from the study. The corresponding cross-layer support required from the MAC layer is also identified and accommodated in the design. The effectiveness of the components of the proposed solution is then comprehensively evaluated through simulation studies.

Publications & Presentations:

• K. Sundaresan and R. Sivakumar,  "Routing in Ad-hoc Networks with MIMO Links." IEEE International Conference on Network Protocols (ICNP), Boston, MA, USA, Nov 2005 (Best Paper Award).
• K. Sundaresan and R. Sivakumar,  "Rate vs. Range vs. Reliability: How to Best Exploit MIMO in Ad-hoc Networks."  Poster Presentation, ACM International Symposium on Mobile Ad Hoc Networking and Computing (MOBIHOC), Urbana-Champaign, IL, USA, May 2005.
• K. Sundaresan and R. Sivakumar, "A Unified MAC Layer Framework for Ad-hoc Networks with Smart Antennas." ACM International Symposium on Mobile Ad hoc Networking and Computing (MOBIHOC), Tokyo, Japan, May 2004.
• K. Sundaresan, R. Sivakumar, M. A. Ingram, and T.-Y. Chang, "A Fair Medium Access Control Protocol for Ad-hoc Networks with MIMO Links." IEEE International Conference on Computer Communications (INFOCOM), Hong Kong, China, March 2004.
• K. Sundaresan and R. Sivakumar, "On the Medium Access Control Problem in Ad-hoc Networks with Smart Antennas.'' Extended Abstract in Proceedings of ACM International Symposium on Mobile Ad Hoc Networking and Computing (MOBIHOC), Annapolis, MD USA, June 2003.

Software Downloads:

People:

• Karthikeyan Sundaresan (Student)
• Raghupathy Sivakumar (Professor)
• Mary Ann Ingram (Professor)

References & Related Work:

Related Work

[1] considers a cellular scenario in which the base-station is equipped with a multi-beamforming antenna, and discusses the improvement in static SDMA/TDMA system capacity on performing dynamic slot assignment. [2] and [3] propose MAC protocols for ad-hoc networks with directional antennas. [4] and [5] address the issue of medium access control in ad-hoc networks with switched beam antennas. However, in [4], the goal of the work is to estimate a lower bound on the overall performance of an ad-hoc network with switched beam antennas , and hence the extent to which the work deals with the MAC protocol is limited to the selection of a simple MAC scheme for the overall goal. [5] considers two MAC schemes and concludes that the simple MAC scheme used in [4] is indeed the more efficient option. [6] and [7] use the directive gain provided by directional antennas for the purpose of range extension and minimization of power consumption respectively. Finally, [8] presents the proportional fairness model for the problem of channelallocation in wireless ad-hoc networks. Though we have designed our protocol in the same framework of PFCR we augment it with several design optimizations that are unique to the MEA environment.

References

1. F. Shad, T. Todd, V. Kezys, and J. Litva, Indoor SDMA capacity using a smart antenna base station, in IEEE ICUPC, 1997, pp. 868-872.
2. Y.-B. Ko, V. Shankarkumar and N. Vaidya, Media access control protocols using directional antennas in ad-hoc networks, Proceedings of IEEE INFOCOM, Tel-Aviv, Israel, March 2000.
3. A. Nasipuri, S. Ye, and R. E. Hiromoto, A MAC protocol for mobile ad hoc networks using directional antennas, in IEEE Wireless Communication and Networking Conference (WCNC), 2000.
4. R. Ramanathan, On the performance of ad hoc networks with beamforming antennas, in ACM MOBIHOC, Oct. 2001.
5. M. Sanchez, T. Giles, and Jens Zander, CSMA/CA with beam forming antennas in multi-hop packet radio networks, in Swedish Workshop on Wireless Ad-hoc Networks, Stockholm, Mar. 2001.
6. Romit Roy Choudhury, Xue Yang, Ram Ramanathan, and Nitin H. Vaidya, Using directional antennas for medium access control in ad hoc networks, in Proc. ACM MOBICOM, Atlanta, 2002.
7. Asis Nasipuri, Kai Li, and Uma Reddy Sappidi, Power consumption and throughput in mobile ad hoc networks using directional antennas, in Proc. IEEE IC3N, Oct. 2002.
8. Thyagarajan Nandagopal, Tae-Eun Kim, Xia Gao, and Vaduvur Bhargavan, Achieving MAC layer fairness in wireless packet networks, Proceed-ings of Mobicomm 2000, Aug 2000.
9. Karthikeyan Sundaresan and Raghupathy Sivakumar, Medium access control protocol for ad-hoc networks with switched-beam antennas, GNAN Research Group Technical Report, April 2003.