A Reputation Based Detection Scheme for Securing P2P Home Networked Appliances in HNASS Mazhar Ul Hassan, Dr. Sohail Abbas, Dr. Arshad Muhammad Email: [email protected] Abstract- We are in the era where computing is embedded fairly

ubiquitously in the environment. As peers are autonomous in nature, no peer will fully trust another peer in a p2p network in terms of data, file or service sharing. There are many types of attacks in P2P network, but our concern is to establish a framework which mitigates some of the security threats in P2p network. As p2p are autonomous in nature they may lie to each other about their services they provide. There is a need for a security scheme which can provide improved security for home appliances within a P2P network context. The main aim of this paper is the development of a distributed scheme able to work within the network, applying Secure Component Composition techniques for the purpose of controlling lying of peers about the services they provide. If peers give false information about services they provide, the framework will take certain action against them. This will improve the security of the overall network. Keywords— Peer-to-Peer Networks; Home Networking; Home Networked Appliances; Home Networked Appliances Security; Composite Security

INTRODUCTION Peer-to-peer (P2P) or ad-hoc networking has become one of the most widely discussed terms in information technology in recent years [1]. The term peer-to-peer or ad-hoc refers to the concept that in a network of equals (peers) using appropriate information and communication systems, two or more individuals are able to impulsively work together without the presence of any fixed infrastructure [2]. P2P networks are often referred to as overlay networks [3-4]. Overlay networks are networks that are built on top of another existing network. Nodes in overlay networks act as connections via virtual links each corresponding to a path in the underlying physical network. In a P2P network, peers can join or leave the system without any intervention from a centralized server, which facilitates seamless integration of any number of new nodes (peers) in the network. The decentralized nature of P2P networks facilitates scalability. P2P networks are interesting in their own right, but in this paper we consider them as a means to facilitate the deployment of networked appliances within the home. We believe overlay networks, and P2P networks in particular offer attractive capabilities for the implementation of a networked appliance framework. In this context it will be useful to consider further the concept of networked appliances. A Networked Appliance is defined as “a dedicated function consumer device with an embedded processor and a network connection” [5]. Networked appliances are gaining interest over time and can be categorized in two different perspectives.

Hardware Perspective: A networked appliance (NA) is defined as a dedicated function consumer device with an embedded processor and a network connection. Software Perspective: A networked appliance (NA) provides independent services that can be discovered and used by other networked appliances to extend the functionality it supports beyond that which it was initially designed to do [6]. As we know that peers can join or leave the system without any intervention from a centralized server, which facilitates seamless integration of any number of new nodes (peers) to existing systems. P2P users usually have no idea about whom they are interacting with. Moreover a given peer may know virtually nothing about other peers. Thus, it becomes extremely difficult for a user to trust another user. The issue of security issues within P2P networks is important, especially in the context of home appliances. For peerto-peer networked appliance systems to be widely accepted and adopted, they must be secure. This paper establishes an in-depth understanding of P2P network security in home appliances, we therefore propose a novel scheme known as the Home Networked Appliances Security Scheme (HNASS) [7]. This scheme has been designed to secure all service requests besides taking measures to protect against any intruders posing threats. The purpose of this scheme is to control peers lying about their services they provide. The remainder of this paper is organized as follows. In order to develop a background understanding of related work, reported research will be discussed and critically analysed, but research shows there are weaknesses within these schemes. Such schemes are discussed in Section 2. In Section 3 HNASS is proposed and explained as a solution to the problem of P2P networked appliance security. Reputation based misbehaviour detection scheme is conducted in Section 4, whereas conclusions and future work are covered in Section 5. RELATED WORK

A number of research initiatives have tried to standardize how devices are interconnected directly to each other and to let them cooperate with each other. We have discussed some of the common standards being used to interconnect networked appliances within the home in [7-9]. These research initiatives are Universal Plug and Play (UPnP), the Open Services Gateway Initiative (OSGi), the Home Electronic System (HES), ePerSpace, MediaNet, VisNet, Semantic HiFi, BETSY, WCAM and RUNES have all tried to standardise how devices are interconnected. All of these research initiatives have tried to standardize how devices are interconnected by describing and discovering services using attribute-based techniques. A related approach is provided through the concept of services by the Networked Appliance

Service Utilization Framework (NASUF). In a service enabled network, appliances offer their services to other appliances when needed. These services are dynamically discovered and composed within a P2P network without any centralization [10]. In other words, all the services and functionalities of a device can be discovered, composed and used by all other devices within the network. NASUF integrates heterogeneous devices, enables seamless communications, and allows services provided by devices to be shared. Within NASUF this integration is achieved using the JXTA protocols [11]. These protocols allow any device to be connected to the network independent of the platform, programming language, or the transport protocols devices implement. These standards address a number of challenges but from our point of view is that security in these frameworks is often addressed in only a limited sense. Authentication – often implemented using passwords – is the primary technique, with an assumption that Java will tackle the other security requirements of the system. The result is that the security mechanisms impose functionality restrictions (e.g. preventing anonymous access to services where security can be guaranteed), and cover only a limited collection of security concerns (e.g. ignoring trust, data flow or wider confidentiality issues). We have therefore looked at layering an additional security system, implemented primarily as a set of services within the network, on top of an existing framework. For our foundation framework we have selected NASUF, since it provides rich service discovery and composition capabilities. Moreover it promotes the decomposition of physical devices into their constituent services, for example a TV publishes the visual, audio and RF-Receiver functions as independent services that can be simultaneously discovered and used within the home environment. Consequently NASUF supports a wide variety of dynamic scenarios, inhabiting a particularly challenging security space. We also have ready access to a NASUF test-bed within our research lab, allowing a more realistic set of scenarios to be tested. Since we have revealed there is generally a lack of security standards in the available P2P networked appliance schemes. Thus there is an urgent need for the development of suitable security protection for networked appliances. We believe the framework proposed in this paper could provide a solution to one of the foremost aspects of such security protection as without security assurance, networked appliances are unlikely to become widely adopted. In the following section, we present our novel scheme to provide security for NASUF. HOME NETWORKED APPLIANCES SECURITY SCHEME The Home Networked Appliances Security Scheme (HNASS) [7] follows an intermediate approach between the existing schemes and some of the new concepts which we propose as a part of this scheme, based on developments in secure service composition. We have developed HNASS in order to provide security for home networked appliances based on the overview process shown in Figure 1.

Fig.1. Home Networked Appliances Security Framework [7-9].

In HNASS all peers outside of our P2P networked appliances network can freely use services; similarly our P2P networked appliances can use services outside its network. In both cases connection between peers need to be established but with the restriction they must first go through certain steps before joining the network. There are three components of HNASS, namely a Scanner, Analyser and a Decision Taking Peer (DTP), which work with the other P2P networked appliances in order to provide security. HNASS defines various operations to perform its task. To understand better how this takes place and the required functions of these services in more detail [7-9]. To implement our proposed HNASS scheme we have two different scenarios. In the first instance an external peer sends a service request to the player for using the services of our local P2P networked appliances. The Player peer will forward the external peers information to the Scanner. The Scanner contains a list of peers’ User ID’s. The Scanner will process the information and checks the external peer’s User ID in its black list. Black list is a list of those peers that have been identified as a security threat to the network by the Scanner. If the Scanner finds the external peer’s UID in its black list, it will pass the information to the DTP. In this case the DTP will process this black listed UID and will take a decision to block the external peer’s request for using services. The decision is sent to the scanner so that the scanner can update itself. Figure 2 shows how an external peer sends a service request to the player for using services.

Fig.2. External peer sends a request for using services

In the second case, if the scanner does not find external peer’s UID in its black list it will send a request to the neighbours of the external peers for its trust value. The scanner will receive a trust value (if known) from the neighbours of the external peer. The scanner will pass both the external peer’s UID and its trust to the Analyser. The Analyser will analyse the information and will make a decision about whether to allow or deny the service request. The decision will then be forwarded to the DTP to take action. Here the Analyser updates the scanner periodically about the connected peers because if the external peer’s request is denied by the analyser, the scanner will update its black list. The communication takes place between different devices in NASUF is done by the JXTA protocols used by NASUF. The central idea behind JXTA is the concept of a service. NASUF integrates heterogeneous devices; enables seamless communications; and allows services provided by devices to be shared. JXTA allows any device to discover and communicate with others present in the JXTA P2P network. In JXTA devices are known as peers that sit inside the network. Communications take place between peers, which may reside within and across different networks, by sending XML messages along communication channels called pipes. Pipes are one of the main mechanisms for sending messages between devices, supporting both asynchronous and unidirectional communications. REPUTATION BASED MISBEHAVIOUR DETECTION The reputation system is responsible for the detection of misbehaving peers. By misbehaving peers we mean that a peer that falsely offers DVD quality videos to the external P2P world; however, in reality, the videos do not offer the claimed quality i.e. after downloading and watching the video might turn out to be camera print videos. Another example of a misbehaving peer is that it may give bogus labels to its shared videos that do not match with the original contents; for example, a peer shares corrupted videos while putting decent names to those video files. In our scheme each peer maintains two tables: a firsthand information table and a reputation rating table. Any direct interaction experiences regarding other peers are captured and stored in the firsthand information table which will be periodically shared with the neighbours. The firsthand information table contains the address of a peer and its firsthand reputation rating. In

order to keep the reputation secret, each peer will share only its firsthand reputation information with its neighbours. The reputation table maintains the overall reputation of peers in which reputation is computed from both firsthand and secondhand ratings. We adopt Buchegger et al. [12] model for reputation formulation and detection, which is as follows. Handling firsthand information: We define two variables α and β, the former represents bad behaviour and the latter represents good behaviour; these variables are increased or decreased based on the observed behaviour of a peer. The purpose of the Eq. 1 is to provide a reputation system that incorporates firsthand, secondhand and fading updates. Based on the observed behaviour the variables α and β are updated accordingly as follows. Eq.1 Here a represents bad behaviour (such as sharing a bad file on a good name) committed by a misbehaving peer, b is good behaviour, ζ is the fading factor that fades reputations after a fading timeout in order to assign higher weight to the recent activities, its value falls in [0, 1], and ω is the weight assigned to the secondhand information, its value should be less than 1 which falls in [0, 1]. Reputations i.e. α and β in Eq. 1 will be updated by three processes as depicted in Table 1; whereas the initial reputation rating of a peer will be . Each process will provide different values for a, b, ω and ζ. These updates are explained below. Firsthand update: in this process a and b represent a single direct event observed (1 indicates the confirmation of the observed behaviour, i.e. forward or drop of a packet. These values are simply added to the overall reputation rating, as shown in Eq. 1. Secondhand update: each peer will share its direct experiences, i.e. α and β, of peers in the network. For example, after direct experience with peer j, peer i will share α and β of peer j with other peers, let’s say peer k. After receiving this information shared by peer i about peer j, peer k will treat α and β as a and b in Eq. 1 and will apply the secondhand weight as well, as shown in Table 1. Fading update: reputations are continually faded in order to motivate peers for cooperation or to reduce a chance for a peer that uses its high reputation for malicious activities. As shown in Table 1, the secondhand information will be ignored when reputations are faded. Handling secondhand information: Each peer periodically broadcasts its firsthand information table after PT (publishing timeout), in order to inform its one-hop neighbours about its direct experiences. After receiving firsthand information from a peer, which will become the secondhand information for the receiving peers, the reputation system conducts a deviation test on each individual rating. In other words, the shared ratings are checked against the recorded reputation ratings at the receiving peer, if other peers’ experiences deviate too much from the peer’s own experience, the secondhand rating will not be accepted; otherwise the reputation rating will be updated. Table 1: Reputation update processes S. No.

Process Description

a

b

ζ

ω

1

2 3

Firsthand Update

0/1

0/1

1

1

Secondhand Update

α

β

1

Secondhand Weight

Fading Update

0

0

Fading Weight

1

Detection: in order to setup a criterion for the detection of misbehaving peers, we have to setup a threshold to detect misbehaving peers because the distinction between good and misbehaving peers is very important. We use the following formula for the misbehaviour threshold. Eq.2 After calculating the reputation of a peer, the reputation is checked against the misbehaviour threshold, as shown in the Eq. 2 above. Reputations below the misbehaviour threshold indicate wellbehaved peers and these peers are therefore provided with services, whereas reputations above the misbehaviour threshold are deemed to be misbehaving and the identity of these peers is exchanged with neighbours. Fig 3 shows a tick mark symbol at the right side corner of the screen which shows that the external peer is a well-behaved peer and the analyser has made a decision to allow it. Once the decision is made and a connection is established, the Scanner will monitor the behaviour of the external peer throughout its connectivity inside the network as well as its behaviour with its own neighbours.

Fig. 4. Request of Bad Behaviour Peer is not allowed

As we can see from the experiments above, we are able to successfully monitor the network if its services are being used by external peers. HNASS has successfully scanned and analysed the external peer with its UIDs, properties and its connection with other peers. In this experiment we have also noticed that after the analysis process, the analyser peer has made a decision on the basis of output generated by the scanner to allow the external peer for using services on the network. CONCLUSIONS AND FUTURE WORK Networks are always under security threats, so it is a good assistance for administrator to makeup quick understanding and observation of safety situation before a network comes under attack. In this paper we have proposed a novel scheme to secure home networked appliances using peer to peer networking. HNASS utilizes a combination of three components to provide secure communication between various peers. In addition HNASS introduces measures to protect against attack by an intruder of peers associated with the network. We have successfully scanned and analyzed audio, video and player devices on the basis of their UIDs, properties and their connection with other connected peers inside and outside the local network. In future work we will be conducting research experiments to monitor our proposed scheme for performance across a diverse range of environments. We will evaluate this scheme in HNASS in the near future.

Fig. 3. Request of Good Behaviour Peer is allowed

Fig 4 shows a cross mark symbol at the right side corner of the screen which shows that the external peer has been rejected because of its bad behaviour. Once the decision is made, the Scanner will add a rejected peer into its blacklist, so that in the future it is recognized as a bad peer. The information about this bad peer is also exchanged with neighbours.

REFERENCES [1] [2] [3]

H. Bakht, "Understanding Mobile Ad-hoc Network," Computing Unplugged Magazine, 2004. D. Schoder, et al., "Core Concepts in Peer-to-Peer Networking," in Peer-toPeer Computing: The Evolution of a Distruptive Technology, ed: IGI Global, 2005, p. 308. K. Lua, et al., "A Survey and Comparison of Peer-to-Peer Overlay Network Schemes," IEEE Communications Survey and Tutorial, vol. 7, pp. 72-93, 2005.

[4] [5] [6] [7] [8]

[9]

Z. Li and P. Mohapatra, "On investigating overlay service topologies," Computer Networks, vol. 51, pp. 54-68, 2007. S. Moyer, et al., "Service portability of networked appliances," Piscataway, NJ, USA, 2001, pp. 13-19. A. Mingkhwan, et al., "Dynamic service composition in home appliance networks," Multimedia Tools and Applications, vol. 29, pp. 257-284, 2006. M. Ul Hassan, et al., "A Security Scheme for Home Networked Appliances," in Proceedings of the Layered Assurance Workshop (LAW2010), Austin, Texas, USA, 2010. M. Ul Hassan, et al., "A Secure Composition Scheme for Home Networked Appliances in Peer-to-Peer Networks," in presented at IEEE International Conference on Computer Networks and Information Technology (ICCNIT 2011), University of Peshawar, Pakistan., 2011. M. Ul Hassan, et al., "Secure Service Composition for P2P Home Networked Appliances," in The 12th Annual Conference on the Convergence of

[10] [11] [12]

Telecommunications, Networking & Broadcasting (PGNet 2011), Liverpool, UK, 2011. P. Fergus, et al., "A semantic Framework for self-adaptive networked appliances," in IEEE Consumer Communications and Networking Conference(CCNC'05), Las Vegas, Nevada, USA, 2005, pp. 229-234. S. Microsystems. JXTA v2.3.x: Java Programmer's Guide, 2006. Available: https://labs.oracle.com/spotlight/misc/jxta.pdf S. Buchegger and J.-Y. L. Boudec, "The Effect of Rumor Spreading in Reputation Systems for Mobile Ad-hoc Networks," in Proceedings of WiOpt `03: Modeling and Optimization in Mobile, Ad Hoc and Wireless Networks, France, 2003.